ph255 Lab Manual PDF

PH255: Modern Physics Laboratory
Patrick R. LeClair
August 23, 2012
Contents
1 PH255 Course Syllabus 1
1.1 Course Content & Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Laboratory Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Grading scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Experiments & Schedule 11
2.1 Shorter experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Longer experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3 The Care & Feeding of Laboratory Notebooks 15
4 Laboratory Safety 17
4.1 General rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2 High Voltage Safety & Electronic Equipment Considerations . . . . . . . . . . . . . 17
4.3 General Precautions When Using Lasers . . . . . . . . . . . . . . . . . . . . . . . . 20
4.4 Radiation Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.5 Cryogenics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.6 What to do if an Injury Incident Occurs . . . . . . . . . . . . . . . . . . . . . . . . 22
5 Counting Statistics 25
5.1 Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.3 Binomial, Poisson, and Gaussian Statistics . . . . . . . . . . . . . . . . . . . . . . . 26
5.4 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.5 Preparatory Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.6 Relevant Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.7 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.8 Suggested procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.9 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.10 Discussion and topics for your report . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.11 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
ii
CONTENTS iii
Appendix 1: A Quick Review of Standard Deviation . . . . . . . . . . . . . . . . . . . . . 35
Appendix 2:
137
Cs Decay and Gamma Emission . . . . . . . . . . . . . . . . . . . . . . . 37
6 Gamma ray attenuation 39
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
6.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
6.4 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6.7 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Appendix 1: Electromagnetic spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Appendix 2: Density of various materials . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Appendix 3: Gamma sources available . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Appendix 4: Additional data for
137
Cs and
57
Co . . . . . . . . . . . . . . . . . . . . . . . 50
7 Electron Charge to Mass Ratio e/m 53
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
7.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
7.4 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
7.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
7.7 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Appendix 1: Example Data Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
8 The Photoelectric eect 63
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
8.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
8.5 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
8.6 Suggested Setup & Alignment procedure . . . . . . . . . . . . . . . . . . . . . . . . 67
8.7 Experimental Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
8.8 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
8.9 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Appendix 1: UV Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Appendix 2: Visible Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Appendix 3: Lamp and Filter Reference Data . . . . . . . . . . . . . . . . . . . . . . . . . 74
iv CONTENTS
9 Observation of Atomic Spectra 77
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
9.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
9.5 Supplies & Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
9.7 Discussion & Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
9.8 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Appendix 1: Reference Spectral Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Appendix 2: USB-650 Red Tide Specications . . . . . . . . . . . . . . . . . . . . . . . . 97
10 Speed of Light 99
11 Plancks Constant from LEDs 101
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
11.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
11.4 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
11.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Appendix 2: LED Wavelengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
12 Measurement of e/k 111
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
12.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
12.5 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
12.7 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
12.9 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Appendix 1: Simplied 2N1724 Datasheet . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Appendix 2: Simplied ECG-130 Datasheet . . . . . . . . . . . . . . . . . . . . . . . . . . 121
13 Polarization and Diraction of Light 123
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
CONTENTS v
13.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
13.4 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
13.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
13.7 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
14 Fine Structure in Atomic Spectra 135
14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
14.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Appendix 1: Reference Spectral Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Appendix 2: Angular Momentum Correction to Hydrogen Levels . . . . . . . . . . . . . . 144
Appendix 3: Center-of-Mass Correction to the Hydrogen Energies . . . . . . . . . . . . . 145
15 Charge Quantization 151
15.1 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
15.4 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Appendix 1: Viscosity of dry air as a function of temperature . . . . . . . . . . . . . . . . 156
Appendix 2: Derivation of Stokes law and the Navier-Stokes Equation: A crash-course
in uids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
16 Gamma ray spectroscopy 179
16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
16.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
16.4 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
16.6 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
16.7 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Appendix 1: Gamma sources available . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
17 Charge Transport 189
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
17.2 Noise in resistive devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
17.3 Signal averaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
vi CONTENTS
17.4 Summary of Noise and Averaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
17.5 Four point probe techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
17.6 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
17.7 Supplies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
17.9 Data analysis and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
17.10 Format of report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Appendix 1: Solving the van der Pauw equations numerically . . . . . . . . . . . . . . . . 210
Appendix 2: Electrical Resistivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
Appendix 3: Measuring Resistive Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Reference Data 227
17.11 Periodic Table of the Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Bibliography 229
1
PH255 Course Syllabus
1.1 Course Content & Description
Experimental work in the topics that form the subject matter of modern physics, including spe-
cial relativity, quantum physics, atomic and nuclear structure, and solid state physics. PH255 can
be taken at the same time or following PH253.
The primary purpose of PH255 is to become acquainted rsthand with some of the phenomena
that provided the empirical impetus for quantum mechanics and special relativity. These include,
for example, the speed of light, the photoelectric eect, atomic emission spectra, semiconductor
physics, and nuclear decay and detection.
1.1.1 Course Topics
The course will consist of performing and analyzing most of the following experiments:
Counting statistics (scintillation detector)
Electron charge-mass ratio
Determination of Plancks constant
Solar spectra
Polarization & diraction
Millikans electron charge determination
Advanced optical spectroscopy
Atomic spectroscopy
Photoelectric eect
Electrical e/k
B
determination
Gamma ray attenuation
Gamma ray spectroscopy
Speed of light determination
Hall eect & superconductivity
Muon physics
1.1.2 Course Goals and Objectives
General Learning Outcomes for 100- and 200-level courses:
1. Recognizing physics concepts that involve developing mathematical models of ordinary phe-
nomena, such as weights and measures, moving objects and forces. [knowledge, evaluation,
analysis]
2. Knowing the scientic method and the process of critically evaluating scientic information.
[knowledge, comprehension, evaluation]
Anticipated Learning Outcomes for this Course
1
2 1.2 Reporting
1. Students will become familiar with scholarly and research methods. [knowledge, analysis]
2. Students will be able to critically discriminate between reliable and less reliable information.
[analysis,evaluation]
3. Formulating the solution of a physics problem. [analysis, synthesis]
4. Analyzing the accuracy of a result. [evaluation]
5. Estimating the order of magnitude of a result. [evaluation]
Assessment of Outcomes for 100- and 200-Level Courses
1. Assessment by checklists and rubrics of laboratory reports (oral and written).
2. Assessed by performing laboratory experiments that emphasize the application of the scientic
method.
1.1.3 Course Format
PH255 is a purely laboratory-based course. On some days there may be short (15-30 minute)
instructional lectures, but this will be the exception rather than the rule. Students are expected to
read the relevant sections of the laboratory manual before coming to class so they can begin their
experiment immediately upon the start of the class period.
1.2 Reporting
Each of component of your grade is described in more detail below. The relative weight for each
component in determining your overall grade are shown in Table 1.1, and the grading scale used is
detailed in Table 1.2 at the end of this section. There are several types of reports over the entirety
of the course, each described in detail below.
During the rst portion of the course, you will perform a series of 4 shorter experiments designed to
take two lab sessions each. The detailed schedule can be found in Sect. 2. For these 4 experiments,
you will prepare one report on each experiment. Each report must have a dierent format: one
formal written report, one short oral presentation, one short memo, and one research proposal.
These reports are due within 2 weeks of nishing the relevant experiment to avoid a
4% per day late penalty.
During the nal 4 weeks of the semester you will perform one or two two dierent experiments
which are slightly more involved. Some of these experiments will take two full lab periods to set
up and acquire all data, some will take the full 4 weeks. Your group will choose either one two
of the experiments to perform, such that the total time is 4 weeks. For the nal experiments,
your group will give a 30 minute group presentation, to be scheduled during dead week at the
latest. This report can discuss your entire 4 weeks of work or a single experiment, it is your decision.
P. LeClair PH255: Modern Physics Laboratory
1.2 Reporting 3
Table 1.1: Grading Breakdown
Component Sections section (%) total (%)
Short reports Formal written 15
Oral 15
Memo 15
Proposal 15
60
Final reports Oral 20
Summary 5
25
Other Laboratory notebook 10
intro. expt., quiz 5
15
Additionally, the quality of your laboratory notebook will also factor into grade determination.
1.2.1 Formal written reports
One required for the initial shorter experiments. Suggested length: 5-10 pages; no strict limit.
Elaborate write-ups of the experiments are not required, but it is expected that reports will include
detailed graphs and data analysis. Though the format will vary with the experiment and your own
preferences, templates will be provided. A suggested outline is:
Purpose of the experiment
Description of experiment
Theory
Procedure
Data
Sample calculations
Uncertainty analysis
Results and discussion
A good laboratory report should be clear enough that you can read it a few years hence and recall
what you have done and repeat it. It is particularly important that the results be clearly presented
and their signicance and deviation from expected results understood. Grading of formal reports
will be according to the following criteria:
Theoretical and experimental motivation, 10%
PH255: Modern Physics Laboratory P. LeClair
4 1.2 Reporting
Description of experiment, 40%
Analysis of data and results, 40%
Style and quality of written English, 10%
Analysis should include things such as comparisons with theoretical models, statistics and error
analysis, and plots/tables; style includes things such as references, captions for gures and tables,
a short abstract, conclusions, and general proofreading; the description of the experiment need not
reproduce the laboratory procedure, but should provide sucient detail for another experimenter
to reproduce your results. The overall quality of your data will aect both the description and
analysis sections.
1.2.2 Short oral presentations
One required in total for the initial shorter experiments, 15 minutes duration. Following roughly
the outline above, your group will present the outcome of your experiment. Oral presentations must
be scheduled within two weeks of the completion of the experiment, at a non-class time of your
convenience. All group members must present. Grading of oral presentations will be according to
the following criteria:
Theoretical and motivation, 30%
Understanding of experiment and quality of data 40%
Error analysis, interpretation, and conclusions, 20%
Presentation quality, 10%
A few guidelines for each section: motivation and theory should include the fundamentals only,
no long derivations, critical relationships expressed; understanding should show and discuss the
apparatus, calibrations employed, primary (raw) data, and reduced data plots; analysis should
include how uncertainties are derived, systematic and random uncertainties, and comparison with
known values; presentation quality includes timing, clarity of message, and a smoothly-owing nar-
rative.
Other factors under consideration are your loudness and articulation, rate and timing, spontaneity
(e.g., thinking on your feet), eye contact and expression, posture and gestures, professionalism,
visual aids, and responses to questions.
1.2.3 Memos
One required in total for the initial shorter experiments. Suggested length: 3-5 pages; hard limit
of 5 pages in maximum. A concise memorandum summarizing the results of the experiment.
The memorandum is a sort of role-playing exercise: you play the role of an outside contractor,
1.2 Reporting 5
specializing in analytical work, who has been hired by a technology rm to perform a specic
experiment. For example, for the gamma ray attenuation experiment, you might imagine that you
are to investigate the properties of candidate radiation shielding materials. You are writing this
memorandum for a very busy manager at a technology rm who has little time for (or interest
in) esoteric details: you must get to the point quickly, without sacricing critical information. A
template will be provided. Creativity is encouraged, and the format will vary with the experiment
and your own tastes. A suggested outline is:
1. Background: what is the problem addressed?
2. Theory: a very short (2 paragraph) outline of the physics
3. Experimental calibration procedures
4. Results and Discussion
5. Conclusions: did you solve the problem? To what degree?
6. Appendices: important, but secondary data.
The grading of memos will be somewhat between that of formal reports and oral presentations, but
with brevity and precision playing a larger overall role.
Theory and motivation, 10%
Description of problem, 20%
Description of experiment, 20%
Analysis of data, results, and uncertainties, 30%
Relevance of results to the stated problem, 10%
1.2.4 Proposals
One required in total for the initial shorter experiments. Suggested length: 5-10 pages; hard limit
of 15 pages maximum. Another role-playing exercise! You have just completed an experiment,
and are excited by the results. However, exploring them further will require research funding for
student stipends, supplies, equipment, etc. Therefore, you will write a short (fake) proposal to
the National Science Foundation (NSF) introducing the problem at hand, your preliminary results,
and a description of the proposed research and its potential impact.
i
A template will be provided.
Creativity is encouraged, and the format will vary with the experiment and your own tastes. A
suggested outline is:
1. Project summary: single-page summary of the proposed research, including intellectual merit
and broader impact.
2. Project Description
- Introduction to the problem
i
Just so were clear: you are not actually submitting proposals to the NSF. This is just an exercise.
6 1.2 Reporting
- Background theory and prior results by others
- Initial results (i.e., the result of your experiment)
- Proposed research plan (i.e., what more do you want to do, and what do you need to do
it)
- Broader impact of proposed research (who will it aect, student training, industry, . . . )
3. Management plan (rough timeline, personnel involved, management structure)
4. Equipment and other resources (both what is available and what you require further)
The grading of proposals will also be somewhat between that of formal reports and oral presenta-
tions, but with a stronger emphasis on the creativity of the proposed follow-up experiments.
Theory and motivation, 10%
Description of problem, 20%
Description of existing results (including analysis), 20%
Proposed experiments (including experimental feasibility and theory), 30%
Impact of proposed research, management plan, 10%
As with the other types of reports, a template will be provided as a guide. As with the memos,
creativity is encouraged!
1.2.5 Final oral presentation
Just like the short oral presentation, but more so: 30 minutes in duration (5 min). The grading
will proceed according to the guidelines for the short oral presentations. An informal written
summary of your nal projects is also required, details of which will be provided as the semester
unfolds.
1.2.6 Laboratory notebook
You must keep a laboratory notebook as a group. Its purpose is to document your activities in
the laboratory, including experimental data and preliminary plots, procedures you followed, the
location of electronic data, etc. More details will be provided in Sect. 3 and in a subsequent
handout, below are some brief guidelines.
1. Record what you do in such a way that you will understand it a year from now, and could
repeat the experiment.
2. Use a bound book with pages big enough to tape/glue in printed items (e.g., graphs)
3. Put your name and contact information on the rst page.
4. Leave a few blank pages at the beginning for, e.g., a table of contents, contact information
for lab partners, useful constants, etc.
1.3 Laboratory Safety 7
5. Spread out leave the left-hand page blank in case you need it (e.g., analysis, follow-up notes)
6. Enter your notes chronologically and date each page, even if working on several labs at once.
Also note who you are working with.
7. Write down notes before, during, and after you work with equipment. Include schematics or
sketches of the equipment, block diagrams illustrating functional relationships, etc. Do not
take notes on separate pages and copy them in.
8. Record all measurements in raw form, do conversions/calculations as a second step. Always
include units!
9. Do progress checks as you go along quick graphs, analysis, computations. This is a good
way to catch glitches or bad data, and is the essence of experimental work!
10. Before you leave the lab, take the next step by entering your into the computer and doing a
quick graph (if appropriate). Again, this helps you determine if your data is sucient and
assess overall quality before it is too late.
11. Continue to use your lab notebook as you nalize the experiment and write the report.
Updated graphs, error calculations, comparisons with other experimenters (i.e. accepted
values), conclusions, ideas for improvements would be appropriate. If your notebook is done
correctly, a short lab report comes almost directly from your notebook.
1.2.7 General aspects
All reports, of any type, must include proper citations for instance, reference data and images
used in presentations or reports must have their source properly attributed if they are not your
own work. All quantities must include units (SI and derivative units only) and uncertainties where
appropriate. With the exception of the short oral presentations, reports are due one week after
the completion of the experiment, with a two-day grace period. Reports not received in this time
frame will be reduced by one letter grade, with additional reductions for extreme delays. Short
oral presentations must be scheduled within two weeks of the completion of the experiment. The
nal written report is due on the last day of classes, and the nal presentations will be scheduled
during the last week of classes (a.k.a., dead week).
1.3 Laboratory Safety
The safety policies and procedures listed below must be followed in the laboratory. Failure to comply
with these rules may result in dismissal or suspension from the laboratory. More importantly, failure
to comply with these rules will put you at risk of injury!
Be aware of the locations of re extinguishers, rst aid kits, debrillator, and emergency exits.
The emergency telephone is in the hallway (dial 911 for emergencies).
All accidents or injuries, however slight, must immediately be called to the attention of the
laboratory instructor who will advise the student on the course of action to be taken.
8 1.3 Laboratory Safety
Unauthorized experiments are prohibited. Follow the laboratory directions carefully, and ask
the instructor before deviating from them.
Food and beverages are prohibited in the laboratory.
Do not remove any laboratory equipment or supplies from the laboratory.
Never work in the laboratory without supervision.
Eye protection must be worn at all times when performing experiments involving liquid
nitrogen or pressure vessels.
The thermometers used in this laboratory may contain mercury. Mercury is toxic. Mercury
is a silvery liquid metal at room temperature. If the thermometer is broken the contents
may spill out. Avoid skin contact with liquid mercury. Notify the lab instructor in case of a
mercury spill.
The shielding used in some experiments is made of lead. Never handle lead without wearing
gloves, and wash your hands afterward to be on the safe side.
Use liquid nitrogen only according to the written instructions in the laboratory manual. Avoid
skin contact with liquid nitrogen. It is extremely cold and can cause frostbite or serious burns.
Liquid nitrogen is to be used in open Dewar containers only. Never pour liquid nitrogen into
unauthorized containers, especially airtight containers. Liquid nitrogen is a continuously
boiling liquid. If its volume is conned, then the vapor pressure created by the boiling liquid
can build up to dangerous levels. By dangerous, we mean shopping-for-an-eyepatch levels or
worse.
Never look directly into a laser beam or its direct reection or point a laser beam at someone
else.
Use care when handling electrical equipment in order to avoid electrical shock. Do not alter
electrical connections unless the relevant sources are fully powered o.
Do not handle electrical equipment with wet hands.
Notify the instructor in case of damaged power cords or other problems with electrical equip-
ment.
Careless acts are prohibited. You will know a careless act when you think of one.
A more extensive discussion of laboratory safety can be found in Sec. 4.
1.4 Grading scale 9
1.4 Grading scale
Finally, here is the grading scale that will be used to determine letter grades at the end of the
semester.
Table 1.2: Grading Scale
Grade
Letter Numerical Min. % Description
A+ 4.33 97%
Superior ability or attainment signicantly
beyond all minimum expectations
A 4.00 93%
A- 3.67 90%
B+ 3.33 87%
Good ability or attainment which meets and
exceeds many minimum expectations
B 3.00 83%
B- 2.67 80%
C+ 2.33 77%
Ability or attainment which is acceptable and
meets all minimum (required) expectations
C 2.00 73%
C- 1.67 70%
D+ 1.33 67%
Ability or attainment which does not meet all
minimum (required) expectations
D 1.00 63%
D- 0.67 60%
F 0.00 0-59% Not appropriate for a minimum professional
level of performance
10 1.4 Grading scale
2
Experiments & Schedule
2.1 Shorter experiments
Shorter experiments for the beginning of the semester are designed to take two lab periods for setup,
data collection, and cursory analysis. Full analysis and reporting will be carried out as homework
during the following week. Each group will choose four experiments from the list below to perform
during the rst portion of the course:
1. speed of light
2. counting statistics and gamma ray attenuation
3. observation of atomic and thermal spectra
4. electron charge to mass ratio (e/m) and photoelectric eect (h/e)
5. polarization and diraction
6. interferometry
7. absorption spectroscopy
8. Plancks constant and Boltzmanns constant
These will have a variety of reporting requirements, as detailed in the syllabus (Sec. 1.2). Over the
course of these experiments, you will be required to do four reports of dierent types:
1. formal written lab report
2. short memo
3. research proposal
4. group oral presentation
You may choose which type of report you will do for each experiment, keeping in mind that you
must do one of each. You will be given basic templates and more detailed guidelines for each type
of report. There are a few specic items that we expect to see in each report (if applicable):
1. Properly formatted equations, tables, and plots, suitable for Physical Review (see https:
//authors.aps.org/STYLE/). Examples will be provided.
2. Proper SI units for all reported quantities.
3. Correctly propagated uncertainties for all reported quantities.
4. Statistical analysis of your data.
5. Regression analysis / curve tting (e.g., linear or gaussian t).
6. Quantitative comparison with a physical model.
11
12 2.2 Longer experiments
7. Full citations/references for any information included in the report that is not your own
original work. Online sources (e.g., Wikipedia) are acceptable, but should not be a sole
source for any topic.
2.2 Longer experiments
During the nal 4 weeks of the semester you will perform one or two dierent experiments which
are slightly more involved and more inquiry based. Some of these experiments will take two full
lab periods to set up and acquire all data, some will take the full 4 weeks. Your group will choose
either one two of the following experiments to perform, such that the total time is 4 weeks:
1. ne structure in atomic spectra (2 weeks)
2. muon decay (2 weeks)
3. molecular and absorption spectroscopy (2 weeks)
4. superconductivity and giant magnetoresistance (2 weeks)
5. Millikan oil drop (electron charge; 2 weeks)
6. electron and light diraction (2 weeks)
7. Giant Magnetoresistance (GMR; 2 weeks)
8. analog electronics and semiconductor devices (4 weeks)
9. build a new experiment (4 weeks + prior discussions)
10. An experiment of your own design, subject to approval (2 or 4 weeks)
For the nal experiments, your group will give a 30 minute group presentation, to be scheduled
during dead week at the latest. This report can discuss your entire 4 weeks of work or a single
experiment, it is your decision. An informal written summary of your nal projects is also required.
2.3 Schedule
For most of the experiments above, the maximum number of students that can be accommodated
at any time is three. Given that, and the time and space constraints, we have a capacity of 12
students for this course. You will be split into groups of at most three students each. You are
free to choose your own group members, but the groups must remain xed through the semester,
barring any problems.
The shorter experiments will be conducted by all groups following the schedule below. After the
shorter experiments have nished, for the last four weeks of the semester each group will perform
two of the two-week experiments of their choosing in sequence. At any given time during the last
four weeks, most of the possible experiments will be occupied. Therefore, it may not be possible
to accommodate every groups rst choice of experiment, but every eort will be made to grant at
least one of each groups rst request between the two experiments.
2.3 Schedule 13
Week Sect. 1 (M) Sect. 2 (Th) topic reading
1 27 Aug 23 Aug Introduction, statistical analysis Taylor Ch. 1-2
2 10 Sept 30 Aug
Two-week experiment 1
Taylor Ch. 2-3
3 17 Sept 6 Sept Taylor Ch. 4
4 24 Sept 13 Sept
Taylor Ch. 5, 6
5 1 Oct 20 Sept Taylor Ch. 7
6 8 Oct 27 Sept
Taylor Ch. 8
7 15 Oct 11 Oct Taylor Ch. 9
8 22 Oct 18 Oct
TBD
9 29 Oct 25 Oct
10 5 Nov 1 Nov
Final experiments 11 12 Nov 8 Nov
12 19 Nov 15 Nov
13 26 Nov 29 Nov
14 3 Dec 6 Dec Dead week / nal presentations
14 2.3 Schedule
3
The Care & Feeding of Laboratory Notebooks
On the following page, we reproduce an excellent guide from MITs OpenCourseWare on main-
taining a laboratory notebook. (This guide is available under the Creative Commons Attribution-
NonCommercial-ShareAlike 3.0 United States license. See http://ocw.mit.edu/terms/#cc for
terms and conditions.)
15
Requirements for Experimental Notebooks
MIT Department of Physics
(Dated: February 14, 2008)
Learning to maintain a laboratory notebook is one of
the most important skills you will develop in Junior Lab.
A good laboratory notebook is essential when you be-
gin to write papers or to develop oral presentations sum-
marizing your experimental eorts. A clear well-written
narrative that includes experimental schematics, plots of
raw data, and details of your analysis methods will en-
able you to receive quick feedback and assistance during
lab sessions from peers, TAs and section instructors.
A poorly maintained notebook will prove immensely
frustrating to you and your instructor. It is very di-
cult to answer questions like why didnt the experiment
work or why was my result o by an order of magni-
tude? without being able to clearly and easily trace your
eorts using your notebook. Dont count on being able
to recall the gain of your amplier even one day after a
lab session!!! The following is a list of specic guidelines
to follow when performing laboratory work.
1. Create a descriptive table of contents and make an
entry every time you add new material:
Date - Contents - Page Dont
use generic entries like Day 1 or Analysis. In-
stead, produce records of signicant milestones:
e.g. Calibration of NaI Scintillation Counter
Using Ba-133 and Na-22 Check Sources, Plot
of monochromator linearity over the visible spec-
trum, Montecarlo Simulation of Mean Slant Path
Distance in Muon TOF Experiment.
2. Sign and date each page demonstrating authentic-
ity.
3. Dont ever erase, use white-out, or tear out pages
of a lab notebook. Indicate mistakes by simply
drawing a single, neat line through the item. These
may prove to not so incorrect as initially thought
and will very often be useful as a guide to how the
experiment was done and provide clues on how to
better execute the experiment next time.
4. Loose-leaf pages are never acceptable within a lab
notebook. Graphics or tables generated by com-
puter must be neatly taped into the notebook. Re-
member to annotate these types of graphics with
as much information about how they were created
as possible.
5. Preparatory questions and solutions should be writ-
ten in your lab notebooks.
6. Following the preparatory questions, list the ob-
jectives of the experiment. Restate the essential
physics of the experiment in your own words!!!
7. After listing the objectives, identify the things you
will have to do, the data you must obtain and iden-
tify the required calibrations.
8. On the rst day of a new experiment you should
sketch a block diagram of the experimental appa-
ratus..
9. Identify the location of the data les or long anal-
ysis programs if they are too big to directly enter
or tape into your notebook. Analysis scripts, func-
tional forms for non-linear ts, etc. should always
be present in your notebooks.
10. Note typical readings and instrumental settings
so to be able to quickly setup an experiment on
subsequent days.
11. Sketch waveforms at various places within the sig-
nal chain. This will help ensure your understanding
of each component and permit you to rapidly iden-
tify equipment failure.
12. When tabulating data, make neat columns with
headings and always note correct units.
13. Dont wait until after the session has ended to visu-
ally examine the quality of your data. Make prelim-
inary, hand drawn plots of data, AS THEY ARE
ACQUIRED. These initial plots will very often
save you time and frustration in making sure that
your data are reasonable and suggestive of the be-
havior you expect. The importance of making
preliminary plots and analyses in real-time
cannot be overstated.
14. Your notebook should contain diagrams, narra-
tives, tables of raw data, formulas, computations,
reduced data, error analysis and conclusions in a
neat compact, orderly arrangement.
Bring your notebook to every lab session and
to all oral exams. Failure to do so will result in
penalties to your grade!
4
Laboratory Safety
It is required that you read this entire section before doing any PH255 lab work. After reading the
safety information below, please print and sign the lab safety rules acknowledgment form on page 24.
UA safety information and procedures can be found at http://bama.ua.edu/~ehs/.
4.1 General rules
Two primary rules for lab safety are that you never work alone in the lab, and you never operate
equipment without proper training. At least one other person must be present in the lab for you to
work on any experiment or exercise. The instructor will usually be happy to assist you if you need
any extra time in the laboratory beyond normal class hours. Proper training includes reading the
laboratory procedure for a given experiment and asking the instructor for a quick demonstration
of the operation of the equipment involved. A short summary of some of the basic rules:
If you dont know what you are doing WITH CERTAINTY, ask for help.
Dont assume that you can just gure it out, ask for proper instructions.
Make sure you are trained on all equipment and procedures before beginning.
Never work alone.
If you must work alone,see the preceding rule.
If others urge you to follow unsafe practices, refuse.
Taking shortcuts to speed up an experiment is not only potentially dangerous, it usually has
the opposite eect in the end. A hastily done experiment usually needs to be repeated.
If an injury occurs, remain calm, dial 911, and only provide rst aid if you are absolutely
certain of what to do.
Other general rules are to always put things back in their proper place when you are nished, and
leave the lab in a better condition than you found it. This is more than just a common courtesy to
others working in the lab, safety and organization go hand in hand.
4.2 High Voltage Safety & Electronic Equipment Considerations
Some of the exercises and experiments in this course involve the use of high voltage. To ensure your
own safety and to prevent damage to equipment in the lab, certain precautions must be followed.
17
18 4.2 High Voltage Safety & Electronic Equipment Considerations
While OSHA denes high voltage to be anything in excess of 600 V, for the purposes of this
course, we will treat all electrical devices as if they had high voltage present.
Before connecting or disconnecting any cable, make sure that the power supply is in the
STANDBY or OFF mode. NEVER connect or disconnect a cable with power supplied.
Never, ever work alone. If an electrical accident does occur, having a lab partner present may
be the dierence between life and death. We have rst-hand experience in these matters.
Some supplies have capacitors inside which may not be discharged for a considerable amount
of time after powering o the supply. Do not open any electrical equipment unless instructed
to do so and supervised by an instructor.
Always keep one hand in your pocket or behind your back when testing any circuit in which
there may be high voltage present. This will help prevent an electrical shock from sourcing
current through your heart.
Remember that it is current that kills. A good (e.g., sweaty) connection of 6 V across your
body can kill as well as a poor connection of 600 or 6000 V.
Unless specically instructed to do so, do not change the polarity of an applied voltage. Under
no circumstances change the polarity with the high voltage on, as this will damage the power
supply.
Gas discharge tubes (e.g., for atomic spectra) use a 5000 V dc supply. Never insert or remove
a tube while the power is on, and never touch the metal connectors on the ends of the tube
while inserting or removing a tube. Only handle the central glass portion of the tubes. The
tubes may also become hot after operation, let the tubes cool down before attempting to
remove them.
When you have nished your exercise or experiment, power down all electronics (except
computers, they stay on).
Keep all cable ends and connectors o the oor. If a cable ends or connector get stepped on,
it will be damaged, and the mating connector will also be damaged when you attempt to put
them together. Connectors and cables cost anywhere from $10 to $100 each the damage
bill adds up quickly!
Electrical equipment represents the large majority of potential hazards in the lab, particularly in
terms of risk indexed against opportunity. There are many sources of moderate or high voltage, and
many occasions to operate electronic equipment. However, if you follow proper safety procedures,
the risk to you is negligible.
Electricity is potentially dangerous because your body operates via electrical nerve impulses. Even
a small amount of external current can disturb or interrupt such transmissions. The amount of
current is what matters, as shown in the table below.
i
i
Source data from http://www.allaboutcircuits.com/vol_1/chpt_3/4.html. This page has an excellent dis-
cussion of electrical safety.
4.2 High Voltage Safety & Electronic Equipment Considerations 19
Bodily eect dc 60 Hz ac 10 kHz ac
Slight sensation Men = 1.0 mA 0.4 mA 7 mA
felt at hand(s) Women = 0.6 mA 0.3 mA 5 mA
Threshold of perception
Men = 5.2 mA 1.1 mA 12 mA
Women = 3.5 mA 0.7 mA 8 mA
Painful, but voluntary Men = 62 mA 9 mA 55 mA
muscle control maintained Women = 41 mA 6 mA 37 mA
Painful, unable to Men = 76 mA 16 mA 75 mA
let go of wires Women = 51 mA 10.5 mA 50 mA
Severe pain, Men = 90 mA 23 mA 94 mA
diculty breathing Women = 60 mA 15 mA 63 mA
Possible heart brillation Men = 500 mA 100 mA
after 3 seconds Women = 500 mA 100 mA
Keep in mind that these gures are only approximate, and will vary from individual to individual.
However, note that with more than just a few milliamperes, you cant let go of the electrical
source. With good contact, your arm-to-arm resistance is about 1 k. A normal 110 V ac wall
outlet thus gives 110 mA, more than enough to kill you if improperly handled!
If someone in the lab does receive an electrical shock, rst turn o the power to the oending
instrument or pry the victim o with insulated materials (gloves, yardstick, plastic). Apply CPR
if you know how (you have learned CPR, right?), and have someone immediately dial 911.
Some other important suggestions if an accident occurs:
If you dont know what you are doing, GET HELP.
Double check that the power is o, and tape its power switch o so no one else can turn it
on.
Use a plastic-handled screwdriver to short out the device before touching wires or electrodes.
If you need to move the oending wire(s), use a long insulating item, such as a wooden meter
stick.
If you must touch the oending wire(s), keep one hand in your pocket, touch the wire with
the back of your hand rst (higher resistance, no chance of grabbing and not being able to
let go).
Never, EVER work alone.
Be sure all electrical equipment is grounded in at least one place, and do not defeat the third
pin on electrical plugs.
20 4.3 General Precautions When Using Lasers
4.3 General Precautions When Using Lasers
The helium-neon and diode lasers used in the laboratory have a total beam power of <1 mW and
should not be capable of damaging your eyes. However, we will not test this assumption. NEVER
allow a laser beam to enter your eye directly. Take care not to look at the laser beam reected from
a specular (mirror-like) surface. When the beam has been broadened by a lens or diusely reected
(e.g., by paper or a matte-nished surface), it is generally quite safe, since the beam energy density
has been strongly reduced. See http://bama.ua.edu/~ehs/New%20Web/laseprog.htm for further
laser safety guidelines.
4.4 Radiation Safety
The radioactive sources used in this lab are expressly intended for classroom use. They are sealed
sources of very low activity, and according to federal safety standards, do not require special
precautions (for example, it is not necessary to have a lm badge to monitor your exposure).
However, we will treat all radiation sources with proper respect, and assume that they are all
potentially dangerous. Some common sense rules to follow when handling or using radioactive
sources in the lab:
The radioactive sources used in this laboratory are stored in a labeled red box inside a shielded
metal box. WHEN YOU HAVE FINISHED USING A SOURCE, PUT IT BACK. Do not
leave any source lying around the lab unattended.
Do not handle the sources any more than necessary. Before doing an exercise or experiment,
think about how to congure the equipment with the goal in mind of keeping the maximum
practicable distance between yourself and the source. This will help reduce the amount of
time you spend in the proximity of the source.
If you suspect that there is radioactivity where you arent expecting it, there is a survey meter
(Geiger counter) in the lab for checking out this kind of situation.
If you wish to use them, lead bricks are available to provide shielding around the source. This
is primarily to reduce the inuence of background radiation on your experiment, they are not
required for safety. To avoid getting any lead on your hands, use the gloves provided in the
lab while handling the bricks, or any other lead objects in the lab. Wash your hands after
handling lead, even if you wore gloves. Use only painted lead bricks; unpainted and oxidized
bricks will have a white powdery coating, these should not be used.
Never remove a source from the laboratory, even temporarily. The sta are legally responsible
for the sources and must periodically account for their presence and condition.
Keep sources away from your body.
Never bring a radioactive source near your eyes because they are particularly sensitive to
radiation.
Be aware of the sources being used in neighboring experiments.
4.4 Radiation Safety 21
Meticulous care must be taken by all students and sta to insure that every source removed from
the repository be returned immediately after its use. For further information on UA radiation
safety, see http://bama.ua.edu/~ehs/New%20Web/radprog.htm
4.4.1 Exposure and Dosage
An excellent guide on radioactivity and radiation protection can be found at http://pdg.lbl.gov/
1999/radiorpp.pdf. This section is adapted from the MIT OpenCourseWare for course 8.13-14,
Experimental Physics I & II. The safety guide at http://ocw.mit.edu/courses/physics/ under
course 8.13-14 is available under the Creative Commons Attribution-NonCommercial-ShareAlike
3.0 United States license. See http://ocw.mit.edu/terms/#cc for terms and conditions.
Ionizing radiation damages tissue; any exposure should therefore be minimized. The unit of radi-
ation exposure is the rem (roentgen equivalent man). A new unit, called the Sievert (=100 rem)
is recommended by the International Commission on Radiation Units and Measurements (ICRU).
Your inescapable dosage from cosmic rays and other background sources is 360 mrem/yr, which
works out to 4.210
2
mrem/hr. The recommended limit to exposure for a member of the general
public is 100 mrem/yr, averaged over any consecutive ve years. If you follow the lab guidelines,
your exposure will be only a small fraction of the dose you receive from the natural background. A
meter is available for you to check the radiation levels yourself.
Radioactive sources emit three types of radiation: high energy helium nuclei (alpha rays), electrons
(beta rays), or photons (gamma rays). Most of the sources in the lab emit only gamma radiation.
Of the sources which do emit alpha or beta particles, most are enclosed in plastic or metals, which
prevent particulate radiation from escaping. Any exceptions will be obviously noted. Any such
specially marked sources should never be handled. Handling of open alpha- or beta-emitters can
result in dangerous dosages to the skin.
The strength of a radioactive source is measured in curies (Ci). A one-curie source has an activity
of 3.710
10
disintegrations/s. The absorbed dose is a quantity that measures the total energy
absorbed per unit mass; it is measured in rads, where 1 rad=100 erg/g. The equivalent dose is
measured in the units discussed above, the rem. The equivalent dose is derived from the absorbed
dose by multiplying by a radiation weighting factor which is a measure of how damaging a partic-
ular type of radiation is to biological tissue. For photons (gamma rays) and electrons and positrons
(beta particles), the radiation weighting factor is unity; for helium nuclei (alpha particles), it is
20; for protons with energy greater than 2 MeV it is 5; and for neutrons it ranges from 5 to 20,
depending on the energy. When you use the meter in the lab, the readings are in rads, and you
must consider the type of particle when you work out the equivalent dose.
22 4.5 Cryogenics
For gamma rays with energy greater than 1 MeV, a useful approximation is that the equivalent dose
due to a source with an activity of C microcuries is 5.210
4
CE
R
2
mrem/hr, where R is the
distance from the source in meters and E
is the energy of the gamma ray in MeV. For gamma rays

with energy less than 1 MeV, this formula is still approximately true for a full body dose. However,
low-energy gamma rays deposit their energy in a smaller mass of tissue than high-energy gamma
rays and can cause high local doses. For example, the local dose to the hands from handling a
10 KeV source can be up to 25 times the value given by the above formula; hands, however, have a
higher tolerance to radiation than inner organs or eyes.
The protective value of shielding varies drastically with the energy of the photons. The intensity
of a soft X-ray beam of soft (i.e. <1 KeV) can be reduced by many orders of magnitude with
a millimeter of aluminum while 1.2 MeV gamma rays from
60
Co are attenuated by only a factor
of 2 by a lead sheet one-half of an inch thick. The best way to keep your dosage down is to put
distance between you and the source. If you stay a meter away from most sources in the lab, you
will be receiving, even without any lead shielding, a dose which is much less than your allowable
background dose. If, however, you sit reading the write-up with a box of sources a few inches away,
you may momentarily be receiving ten to a hundred times the background level.
4.5 Cryogenics
Liquid nitrogen is chemically inert, but it can cause severe frostbite. Wear gloves and protective
glasses or a face shield when transferring or transporting liquid nitrogen. Do not put liquid nitrogen
in closed containers the boilo of liquid nitrogen will cause pressure to build up until something
explodes.
4.6 What to do if an Injury Incident Occurs
1. Dont panic. The accident scene may be very scary. Resist the urge to rush in without
thinking, or run away.
2. Take charge of yourself and others. Take a few seconds to think and develop a plan. Dial
911. Give rm, clear instructions to bystanders.
3. Approach the victim safely. Turn o any electricity, move the victim away from conductors
with insulating tools (yardstick, gloves). Stabilize falling objects.
4. Perform emergency rst aid if you know what to do. You may do more harm than good if
you do not know the correct procedures. Perform CPR, arrest bleeding if necessary. Do not
move victim if possible.
5. Stay with the victim until help arrives, or delegate a bystander if you must deal with equip-
ment.
A summary of the preceding safety rules:
4.6 What to do if an Injury Incident Occurs 23
If you dont know what you are doing FOR SURE, seek help.
Get proper training for whatever equipment and procedures you encounter, and
DONT assume you are so smart you can derive it all on the spot.
Never work alone if at all avoidable.
If you must work alone, make sure someone knows where you are, what you are doing, and
when you should be done.
If others urge you to follow unsafe practices, refuse loudly and rmly.
It is your body, and you need it for a long time.
If an injury occurs, remain calm, dial 911, and only provide rst aid if you know what to do
FOR SURE.
24 4.6 What to do if an Injury Incident Occurs
LAB SAFETY RULES
You must have a signed copy of this form on le to work in the PH255 laboratory.
The safety policies and procedures listed below must be followed in the laboratory. Failure to
comply with these rules may result in dismissal or suspension from the laboratory.
Be aware of the locations of re extinguishers and rst aid kits.
The emergency telephone is in the hallway (dial 911 for emergencies).
For non-emergency issues dial 8-5454 to contact campus police.
All accidents or injuries, however slight, must immediately be called to the attention of the
laboratory instructor who will advise the student on the course of action to be taken.
Unauthorized experiments are prohibited. Follow the laboratory directions carefully.
Food and beverages are prohibited in the laboratory.
Do not remove any laboratory equipment or supplies from the laboratory.
Never work in the laboratory without supervision.
Never look directly into a laser beam or point a laser beam at someone else.
Use care when handling electrical equipment in order to avoid electrical shock.
Notify the instructor in case of damaged power cords or other problems with electrical equip-
ment.
Careless acts are prohibited.
I have read and understand these rules and agree follow them to the best of my ability.
Student Name:
Student Signature:
CWID:
Course Number:
Date:
5
Counting Statistics
5.1 Hypothesis
Gamma ray emission from a radioactive source occurs as random, independent events. As such,
the distribution of emission times should follow Poisson statistics. For a large number of decays
and a small decay probability, the statistics should therefore be approximately Gaussian.
5.2 Introduction
No matter how many measurements of a quantity we make, we will never know its true value. If
we make a large number of measurements under nominally identical conditions, then the average
of this collection of measurements gives us an estimate of the true value. We might logically expect
that the more measurements we make, the better our estimate of the true value. In some cases, the
underlying statistics of the randomness in the measurements allows us to determine how far our
estimate is from the true value. This is the case for radioactive decay, where the probability for
decay is the same for all identical atoms in the sample of interest. This makes radioactive decay
an example of a sequence of independent random events, in which the occurrence of any event has
no eect on the occurrence of any other. Repeated measurements of independent, random events
occurs often in physics, and the goal of this laboratory is to learn how to analyze such experiments.
Since the decay of an unstable nucleus is entirely random, we cannot predict when a particular
atom will decay. Because it is equally likely to decay at any moment, however, we can predict the
mean number of atoms which have decayed after a prescribed amount of time. If the probability
per unit time of a given atom decaying is , then the mean number of original atoms remaining
after a time t out of an original population of N(0) atoms is governed by an exponential decay:
N(t) = N(0)e
t
(5.1)
Closely related to the decay rate above is the half life of the isotope t
1/2
, the amount of time on
average it takes for half of a given sample of radioactive material to decay. From the exponential
decay, it is clear that
t
1/2
=
ln 2
(5.2)
If a particular isotope has a decay probability of =1 10
5
s
1
, meaning any given atom has a
probability of 1 in 10
5
of decaying during any given second, this implies a half life of t
1/2
=6.9310
4
s
(19.25 hr). The exponential decay is only an approximate solution for real radioactive decay for two
25
26 5.3 Binomial, Poisson, and Gaussian Statistics
primary reasons. First, the exponential function is continuous, while the number of atoms remain-
ing can only take on integer values. Second, since the decay process is random, the exponential
decay is only statistically true. We only expect the exponential behavior to hold for an extremely
large population of atoms. Since we are normally dealing with very large numbers of atoms, of the
order of Avogadros number, the approximation is usually quite good.
The analysis above applies to a particular sort of experiment, one in which we simply measure the
number of non-decayed atoms as a function of time. The present experiment is somewhat dierent,
in that we will make repeated measurements of the number of decay events occurring within a
certain xed amount of time. If the measurement time is small compared to the half life, then
the number of decays observed will be small compared to the total number of atoms present, so
we may assume that the number of non-decayed atoms in the sample is not changing signicantly
over the course of our measurement. On the other hand, the decay rate should be constant in this
case, with the number of decays in a given xed amount of time distributed about a characteristic
mean value. What we wish to establish is that their distribution is governed by the statistics of
independent random events.
How can one judge whether a certain process really has a steady rate on time scales of the experi-
ment? There is only one way: make repeated measurements of the number of counts x
i
in a time
interval t
i
, and see if there is a trend in the successive values of x
i
/t
i
. These ratios are sure to
uctuate, since we are inherently dealing with a random variable, and the question is whether the
observed uctuations are consistent with a xed steady rate. To answer this question, we will need
the expected distributions for a truly random process.
5.3 Binomial, Poisson, and Gaussian Statistics
From the decay probability per unit time and the fact that individual decays are independent
events, we can calculate the probability that out of a sample of N atoms, we measure x decays per
unit time. We do this in the same way we would calculate the odds when throwing dice or tossing
coins, using the binomial distribution. If the probability per unit time of decay is , the probability
per unit time of no decay is 1 . Putting these two facts together, the probability per unit time
of x decays from N atoms is
i
P
(x[N) =
_
N
x
_
x
(1 )
Nx
=
N!
x! (N x)!
x
(1 )
Nx
(5.3)
Though this an exact expression, it is unwieldy and computationally useless given the appearance
of numerous factorials. For a very large collection of atoms and a small decay probability
ii
, we
i
Recall that
_
n
k
_
=n!/k!(n k)!.
ii
Say, of order N 10.
5.3 Binomial, Poisson, and Gaussian Statistics 27
can take the limit of the binomial distribution[1] as N and obtain the somewhat friendlier
Poisson distribution:
P
(x) =

x
x!
e
Poisson distribution (5.4)

Here is the expected number of decays per unit time (the population mean), x the number of
decays per unit time actually observed, giving P
(x) as the probability for measuring x decays per

unit time if are expected. The population mean is related to the decay rate and number of
atoms by =N. The use of the Poisson distribution is justied when the number of events per
unit time is steady over the course of the measurement. On the other hand, if our experimental
data is found to obey Poissonian statistics, we can be reasonably sure that we are observing a
process governed by a steady decay rate.
Though an improvement, the Poisson distribution remains unwieldy and restricted to sets of inte-
gers, owing to the appearance of x!. If the true number of decays per unit timet is large
iii
, the
Poisson (or Binomial) distribution can be approximated by the normal (Gaussian) distribution:
P(x) =
1
2
e
(x)
2
/2
Gaussian distribution (5.5)
This is part of the Central Limit Theorem, one of the most important results in all of statistics.
Crucially, the x in the Gaussian distribution can be treated as a continuous variable if necessary,
while the x in the Poisson distribution must be an integer.
The usual way of writing the normal (Gaussian) distribution is it in terms of the mean which
denes the center of the distribution, and the standard deviation which denes the spread of the
data. The relationship between and for a random process is simply =
, giving
P(x) =
1
2
e
(x)
2
/2
2
Gaussian distribution (5.6)
In the normal distribution, it can be shown that approximately 68% of the observations lie in the
interval [, +], 95.5% in the interval [2, +2], and 99.7% in the interval [3, +3].
Although a true normal distribution is dened in terms of the true mean and the standard devia-
tion , we cannot determine either of these quantities exactly with a nite number of measurements,
we can only measure better and better approximations. In practice, we replace by the observed
mean x, dened as:
x =
1
N
N
i=1
x
i
(5.7)
iii
Say, >100.
28 5.3 Binomial, Poisson, and Gaussian Statistics
where N is the total number of observations and x
i
are the individual observations. We replace
by the sample standard deviation s, dened as:
s =
_
1
N 1
N
i=1
(x
i
x)
2
(5.8)
When only a single observation x is made, the sample standard deviation may be determined by
s =
x. Thus, the mean error for any single result or data point x is

x and the fractional error
is given by 1/
x. This is a very important result for analyzing experimental data. As a practical

point, now you know how to put error bars on a frequency distribution.
The standard deviation is a measure of the random statistical uncertainty in the measurement,
and it becomes part of the experimental error associated with physical measurement. If you have
measured an average count to be 695 with s=26, and another experimenter has measured a count
of 680, then their count agrees with your count, within the experimental error. Nothing is made
of the dierence between the values 695 and 680 because, in all probability, the two results are the
same. A related quantity is the probable error P is dened to be that deviation which is expected
to be exceeded in half of the observations. For a normal distribution, P is given by
P = 0.6745 (5.9)
We can also estimate the relative uncertainty of the measured mean of a number of measurements,
a quantity known as the standard deviation of the mean
x
. This quantity tells you that if you
measure x repeatedly, the sample mean x has an uncertainty
x
compared to the true mean given
by:
x
=

N
(5.10)
For an estimate, can be approximated by s,
x
s/
N. Based on your observations, you can

expect 68% of any measurements of x made in the same way will have a mean of x
x
. Put another
way, if you are primarily interested in the average value of x, then
x
tells you the uncertainty in
that average with respect to repeated measurements
iv
. The standard deviation of the mean is the
typical manner of reporting average quantities with statistical uncertainty:
(best value of x) = x
x
(68% condence) (5.11)
Putting this all together, if we take N measurements for our sample, the frequency with which we
would expect to measure x decays would be, using sample mean and standard deviation:
iv
Remember that the standard deviation tells you that 68% of subsequent measurements will fall within of
the mean x, whereas the standard deviation of the mean tells you that a collection of measurements has a 68% chance
of yielding a mean of x.
5.3 Binomial, Poisson, and Gaussian Statistics 29
frequency(x) =
N
s
2
e
(xx)
2
/2s
2
(5.12)
Frequency here is the number of times that x is the result of your measurement, if you take
N measurements with a mean value of x. The data in Fig. 5.1 below are compiled from 5120
consecutive measurements of the number of 662 keV gamma rays emitted in 40 ms measurement
intervals from a
137
Cs source. The frequency of the measured rates is well-described by a normal
distribution, which can be determined from only the sample average and standard deviation. The
data here have been sorted in to bins of width 2 counts.[1] The binning process just means
that rather than tallying the frequency of exact numbers of counts, we would tally the frequency
of observations falling within xed-width ranges, known as bins. In this case, we counted the
frequency of observations falling in the range [60, 62], [62, 64], etc. Another way of viewing the
binning process is as a sort of rounding for a bin width of 2, we round downward to the nearest
multiple of 2. A bin width of 5 would mean rounding down to the nearest multiple of 5, and a bin
width of 1 just means rounding downward to the nearest integer.
v
0 50 100 150
0
100
200
300
400
500

F
r
e
q
u
e
n
c
y
Gamma counts in 40 ms
40ms dwell
Gaussian fit
x
c
= 75.46 0.12 (68%)
= 8.48
N = 5120
Figure 5.1: Gaussian t to data with an average of x=75.5 counts in 40 ms. The sample standard deviation is s=8.48 counts,
in good agreement with

x = 8.7. The sample standard deviation of the mean is s
x
= 0.119, so our average count rate is
reported as x=75.50.12. (The counts are sorted in to bins of width 2 for this analysis.)
v
Of course, this rounding process is only for constructing the frequency distribution plot, we do not actually
round the data that would be throwing away data! We merely sort it and construct a frequency table based on
intervals determined by the bin width. If we actually rounded it, we could never choose a smaller bin size after that
. . .
30 5.4 Objective
5.4 Objective
In this experiment you will explore the statistics of random, independent events in physical measure-
ments. The random events used in this study will be pulses from a scintillation detector exposed to
gamma rays from a radioactive
137
Cs source. The
137
Cs parent nuclei undergo beta decay to create
excited
137
Ba daughter nuclei, which relax to their ground state by emitting 662 keV gamma rays.
The gamma ray emission from the excited state of
137
Ba occurs randomly, as does the decay of its
parent isotope,
137
Cs. Since the 30 year half-life of
137
Cs is continually replenishing the short-lived
137
Ba, this isotopes activity will be nearly constant for far longer than a lab period, and will suce
for a few hours of measurement.
The scintillation detector software control will allow a measure the source activity 1,024 times for
a specied duration known as the dwell time. From these measurements, you can construct a
frequency histogram and compare with Gaussian or Poisson distributions. Dwell times (i.e., the
measurement time interval) can be changed to change the average number of counts with the dwell
period for further comparison with theoretical predictions. The overall goal is to become familiar
with the statistical analysis of random uncertainties.
5.5 Preparatory Questions
Include your responses to these questions in your report.
1. Describe how a scintillation counter works, starting from the entrance of an energetic charged
particle or photon, and ending with an electrical pulse at the output of the photomultiplier.
Why are subsequent signals independent?
2. Suppose the mean counting rate of a certain detector of random events is 2.1 counts per
second. What is the probability of obtaining zero counts in a one-second counting interval?
What is the most likely interval between successive pulses?
5.6 Relevant Reading
Taylor[1], Ch. 4, 5, 11
Pfeer & Nir[2], Ch. 5.3.1-3, 5.5.1
5.7 Supplies
1. Spectech UCS30 spectrometer
2. NaI(Tl) detector
3. Multi-channel analyzer (MCA) / control electronics
5.8 Suggested procedure 31
4. Pb cylinder for shielding
5. Pb brick
6.
137
Cs source (check out from locked cabinet)
7. detector holder and shelf housing
8. lab notebook and USB drive for saving spectra
Pb brick
Pb cylinder
detector
detector
housing
sources
control electronics
calipers
Pb, Al, Cu
sheets
Figure 5.2: The Spectech UCS30 spectrometer and accessories
5.8 Suggested procedure
5.8.1 Startup
1. Turn on the power to the Spectech UCS30
2. Run the program UCS30 from the shortcut on the desktop
3. Place the
137
Cs source on top of the Pb brick, and place the Pb cylinder around it
4. Place the cylindrical NaI detector inside the Pb cylinder
5. Run the energy auto-calibration from Settings/Energy Calibration/Auto Calibrate - this
will take several minutes. The nal spectrum should look like the one in Fig. 5.3.
6. Once the calibration is nished, you are ready to set up the experiment and begin taking
data.
5.8.2 Setting up Pulse Height Analysis (PHA) mode
1. Under the Mode menu, select PHA (preamp in). This should be the default value.
2. Open the detector settings (Settings, Amp/HV/ADC).
3. Adjust the lower limit of detection (LLD) and upper limit of detection (ULD) to select only
the 662 keV gamma peak. The values are in percent of the maximum energy; about 58 and
71% for LLD and ULD should work well (the peak should be at about 65%).
32 5.8 Suggested procedure
0 200 400 600 800
10
100
1k
10k

137
Cs
I
n
t
e
n
s
i
t
y

(
c
o
u
n
t
s
)
E (keV)
660 keV
Photopeak
478 keV
Compton edge
~210 keV
Backscatter
~30, 70-90 keV
X-rays
Figure 5.3: Gamma spectrum of
137
Cs. Aside from the main photopeak at 660 keV, clear Compton edge and backscatter
features are visible, as well as X-ray absorption edges. Note the logarithmic scale on the vertical axis.[3]
4. Run a spectrum using the go button and verify that you are only scanning the 662 keV
peak.
5. Under the Mode menu, select "MCS (Internal).
6. Under Settings/MCS/Dwell, choose a dwell time. This is the time interval over which the
counts within the detection limits are measured.
5.8.3 Running the experiment
Once a dwell time is selected, running a spectra (with the go button) will make 1024 measure-
ments of the total number of counts within the detection region, with each measurement as long as
the specied dwell time. Do this, and save the resulting data (name it sensibly!). Be sure to save
the le in comma-separated value (CSV) format so you can read it in, e.g., Excel later!
In this experiment, you will want to vary the dwell time such that you have a total number
of counts varying by a factor of 5-10. The goal is to accumulate data at three dierent dwell
times, corresponding to three dierent mean counts per dwell time, such that the total amount of
acquisition time at any dwell time is the same. Thus, if your rst spectra is at a dwell time of
200 ms, for 1024 points your acquisition time is 204.8 s. Your second data set at, e.g., 100 ms would
require two separate spectra to have the same amount of total acquisition time, since 1024 points
at 100 ms corresponds to 102.4 s, or half the measurement time as your rst spectra. Though you
are free to choose your own parameters, a suggested plan would be
5.9 Data analysis 33
1 spectrum at 200 ms dwell time
2 spectra at 100 ms dwell time
5 spectra at 40 ms dwell time
During your data analysis, you will combine the 5 spectra at 40 ms dwell time to make one data
set which can be compared with the single spectra acquired at 200 ms dwell time. It is important
that you run the spectra for each dwell time in order and in quick succession. The total acquisition
time, once the experiment is set up, should take one hour or less, which should leave time to begin
analysis during the laboratory period. The data analysis for this experiment is longer than average.
However, the data acquisition itself is more straightforward than usual.
5.9 Data analysis
Though it is somewhat cumbersome for our purposes, we suggest that you use MS Excel for your
analysis, though you are free to use any statistical/analysis software package you are comfortable
with (e.g., Originlab, Sigmaplot, etc.). If you choose to use Excel, a template will be provided to
help with the somewhat tedious task of generating a histogram; see the PH255 web page.
For each of your runs with dierent dwell times, import all of the data (perhaps with a new work-
sheet or le for each dwell time to avoid confusion). If you have 5 measurements at a dwell time of
40 ms, for instance, import all 5 data sets and paste them sequentially into a single set of columns
to make one large data set.
vi
If you have followed the procedure correctly, the product of dwell time
and the number of points should be a constant for each cumulative data set (e.g., 5 sets giving 5120
points at 40 ms, 2 sets giving 2048 points at 100 ms, and one set giving 1024 points at 200 ms, all
giving 204.8 s acquisition time). A cursory qualitative analysis can be performed by simply plotting
the number of counts versus the acquisition time or measurement number. Does the data seem to
be evenly distributed about a mean, or is a trend visible? Indicate the mean in your graph as a
horizontal line.
As a rst level of quantitative analysis, you should do the following:
1. Compute the sample mean x and sample standard deviation s. Both are built-in functions in
Excel.
2. Compare s to calculated by =
with x. Calculate and comment on the numerical

dierence between s and , e.g., which is larger, and why.
3. Compute P, Eq. 5.9.
4. Count the number of times that the deviation [xx[ is greater than and show that this
occurs for approximately 1/3 (strictly, 31.7%) of the observations. (A column of [xx[ and
the function COUNTIF will be useful in Excel.)
vi
When pasting in the sets, be sure to add to the time column so it increases.
34 5.10 Discussion and topics for your report
5. Show that the deviation [xx[ is greater than P approximately 50% of the time.
Next, you should determine whether the count rate (or equivalently, the decay probability) is
constant quantitatively. For each dwell time make a new column to nd the cumulative average,
dened as
r
c
(j) =
i=j
i=1
x
i
i=j
i=1
t
i
(5.13)
where x
i
is the number of counts detected in time t
i
. Note that the sum in the denominator is
just the time at which data point j was recorded after the start of the measurement, which is j
times the dwell time (or the rst column in your data le), and the numerator is the sum of the
data from the rst point through point j. For a truly random process with a steady mean rate ,
r
c
(j) should converge to in the asymptotic limit. Plot r
c
(j) and demonstrate that the cumulative
average converges, and thus that the decay rate is constant over the course of the measurement.
vii
Include error bars to demonstrate convergence.
viii
If you are satised that the decay probability is constant, construct a histogram of the number of
times that a given x was observed versus x (similar to Fig. 5.1. It may be convenient to distribute
the data into xed-width bins as described above, and as demonstrated in the Excel template. Fit
a Gaussian distribution to your data. If you use x as an estimate of and s as an estimate of , a
calculated Gaussian curve with the proper normalization condition should t your data reasonably
well (the normalization for your distributions can be found from the total number of readings).
Comment on the accuracy of the t.
5.10 Discussion and topics for your report
Are the decay rates consistent with the dwell times chosen? That is, do the rates scale with the
decay time properly? Are the standard deviations of the rates for dierent dwell times related as
expected?
For each dwell time, what are the 68% condence limits for the number of counts?
vii
A way to double-check that the distribution is random about the mean is to calculate the reduced deviation, or
z score: z(xi) = (xi x) /. Calculate z(xi) in a new column. If the deviations are random, then the sum of this
column should be very close to zero.
viii
Recall that the error bar for a given point is just the square root of the number of counts, if the decay process is
truly random.
5.11 Format of report 35
Should you use the standard deviation
N
or the standard deviation of the mean
N
in comparing
dierent data sets?
For the observed distribution with the lowest mean rate, take the highest deviation from that mean
and test whether you would be justied in concluding that the detector was malfunctioning at that
time. Do not forget to account for how many chances there were for such a deviation to occur . . .
5.11 Format of report
You may choose one of the following formats for your report:
1. two-page memo (excluding required plots)
Further details and templates are provided for each of these formats. Keep in mind that over the
course of all of the one-week experiments you must do each type of report.
A Quick Review of Standard Deviation
Suppose a series of measurements is made of the value of some unknown quantity. Usually these
measured values will not all be the same. A statistical analysis of the measured values estimates the
quantity and its variability. Analysis of the uncertainty determines the probability that the true
value lies within a certain range. Of course, the percent dierence between two measured values
gives some idea of the range of measured values to be expected, but this is not a very reliable indi-
cator. Whenever a measurement is reported, a determination of the reliability of a measurement is
equally important to report. The mathematical methods used to determine statistical uncertainty
are commonly referred to as error analysis.
A mathematically complete treatment of error analysis is beyond the scope of this document, and
is a part of your required reading[1]. It is necessary to understand the basic methods of error
analysis to properly report the results of the experiments. Some basic assumptions are necessary
to estimate the uncertainties encountered in the measurements and analysis of data. First, it is
assumed that dierences in measurements are due to small random uctuations that are just as
likely to make the measurement higher as it is to make it lower. In some cases there is a systematic
error which always makes a measurement smaller or larger than the true value. Examples of
systematic errors include parallax in reading a meter stick, friction in balance or meter bearings,
tightening of a micrometer screw too much, failure to account for air resistance, etc.
36 5.11 Format of report
In well-designed experiments, systematic errors are accounted for, noted and measured. Under
these conditions, a very large number of measurements of the same quantity should distribute
themselves symmetrically about the simple arithmetic mean or average, which is the best value
of the quantity. The expected variations of the measurements can be described by a quantity called
the standard deviation
The standard deviation is computed in a straightforward manner. Suppose the quantity x is
measured n times. The measured values are labelled x
1
, x
2
, . . . x
n
. First, we calculate the mean,
or average of all the values, denoted x. This is just as you would expect:
x =
1
n
n
i
x
i
(5.14)
Next, calculate the deviation of each measurement from the mean, x
i
x and square the result:
(x
i
x)
2
. We square the result so that deviations above and below the mean dont cancel each
other out, but add to the overall deviation. Finally, add the squared deviations together, divide by
the number of measurements n and take the square root of the result:
=
_
1
n
n
i=1
(x
i
x)
2
(5.15)
A sneaky, and somewhat less tedious formula is given by
=
1
n
_
n
_
n
i=1
x
2
i
_
_
n
i=1
x
i
_
2
(5.16)
A large standard deviation indicates that the data points are spread far from the mean and a
small standard deviation indicates that they are clustered closely around the mean. The reported
standard deviation of a group of repeated measurements should give the precision of those mea-
surements. When deciding whether measurements agree with a theoretical prediction, the standard
deviation of those measurements is of crucial importance: if the mean of the measurements is too
far away from the prediction (with the distance measured in standard deviations), then the theory
being tested probably needs to be revised. This makes sense since they fall outside the range of
values that could reasonably be expected to occur if the prediction were correct and the standard
deviation appropriately quantied.
According to sampling theory, there is a 68% probability that any additional measurement made
of the quantity x will lie within of the mean and a 95% probability that it will lie within 2
of the mean. in most of the experiments of this course, repeated measurements are performed
about ve or ten times. Using the above analysis for less than ve independent measurements of a
quantity is generally not considered to be statistically reliable.
The statistical treatment assumes that all errors are random and ignores any systematic error, such
as improper meter calibration, wind resistance, etc. which might be present. In any laboratory
situation, it is the responsibility of the experimenter to determine whether or not these systematic
errors are signicant and to include them in the estimate of accuracy if they are. On occasion, it
will not be feasible to repeat an experiment several times and nd a mean and standard deviation.
It may be possible, however, to determine or estimate the uncertainties in the measurements of
the quantities used in computing the nal result. For example, a manual measurement of a time
interval is accurate to 0.1 second, due to reaction times in starting and stopping the timer.
137
Cs Decay and Gamma Emission
137
55
Cs
137
56
Ba
7/2 +
11/2
3/2 +
93.5%
6.5%
137
56
Ba
30 yr
2.55 m
stable
1176 keV
0.662 keV
0 keV
Figure 5.4: Energy level diagram showing the decay of
137
Cs to
137
Ba. The
137
Cs decays via two paths, 93.5% follow a
two-step process resulting in the emission of a 662 keV gamma ray. The decay of the Ba nucleus from its excited state is an
internal transition, and contributes to the decay percentages but not the gamma decay factor.[4]
A
137
Cs nucleus can decay via two routes to the
137
Ba ground state, shown schematically in Fig. 5.4.
First, a single beta emission of 1.176 MeV can occur with no subsequent gamma emission, bringing
the
137
Cs directly to the ground state. However, this single-step process occurs for only about
6.5% of all nuclei. The other 93.5% decay via two-step process. First, a 514 keV beta is emitted,
resulting in metastable and short-lived
137
Ba
. The metastable
137
Ba
subsequently (2.55 min half

life) decays to the
137
Ba ground state by emission of a 662 keV gamma.
Though
137
Cs has only a single gamma emission
ix
, there are x-ray emissions that can be observed
in the energy range under study. We list a few of the higher energy emissions in Table 5.1.
ix
Strictly speaking, there is a 283.53 keV gamma, but its intensity is about 10
5
times smaller than the 662 keV
peak.
Table 5.1: X-rays from
137
Cs
E (keV) I (%) Assignment E (keV) I (%) Assignment
31.45 0.00026 Ba K
3
36.38 0.68 Ba K
1
31.72 2.04 Ba K
2
36.65 0.0079 Ba K
5
32.19 3.76 Ba K
1
37.26 0.215 Ba K
2
36.30 0.35 Ba K
3
37.35 0.0481 Ba K
4
6
Gamma ray attenuation
6.1 Introduction
The term gamma ray (or -ray) is an approximate classication for photons of energies E
100 keV10 MeV (f 10
19
10
24
Hz; see Fig. 6.7 in the Appendix). When gamma rays pass through
matter, their intensity may be attenuated by interactions with the matter just as with any other
photons. Gamma rays have three primary mechanisms of interaction: photoelectric eect, Compton
scattering, and pair production. If gamma rays are emitted in a narrow beam toward a block of
intervening matter, these three interactions cause a loss of intensity along the beam direction either
by deection along a dierent direction or absorption. These processes are shown schematically in
Fig. 6.1 below.
crystal
-
+
Compton
Pair production
-
E > 1022 keV
511 keV
511 keV
-
absorption
Figure 6.1: Gamma rays can interact with crystals in a number of ways: Compton scattering, pair production, absorption,
and photoeect (not shown). The combination of these eects is a dispersion in energy, as well as a deection of many gamma
rays from the original beam direction.
Thus, the interaction of high-energy X-rays and rays involves three dierent mechanisms, which
we describe briey below:
1. Photoelectric eect: the photons in the incident radiation are of high enough energy to
release electrons from atoms and molecules in the absorbing or detecting material. Photons
in the beam are thus annihilated, leading to a corresponding reduction in intensity. This
photoeect will lead to a peak in the energy spectrum, known as the photopeak or full
absorption peak. This corresponds to gamma rays depositing all of their energy in the detector
material. The energy of the gamma rays can be determined from the photopeak energy and
compared to an energy level diagram for the parent nucleus and its decay products.
39
40 6.1 Introduction
2. Compton eect: when a beam of high-frequency electromagnetic radiation passes through
a material containing free electrons, an interaction takes place between the incident photons
and the free electrons. In this interaction, photon-electron scattering, energy and momentum
are transferred from the photons in the incident beam to the electrons. X-ray and -ray
energies are rather large compared to the binding energies of the outermost electrons of any
atom, such that the outermost electrons can be treated as essentially free. As a result of
energy transfer to the electrons in the absorbing material, both intensity and energy of the
beam is reduced.
3. Pair production: When a positron and electron collide, they annihilate each other and
produce photons. Energy conservation dictates that the total energy of the emitted photons
equal that of the annihilated particles. Given that the rest energy of an electron or positron
is 511 keV, the energy of the emitted photons must be at least 1.022 MeV. When high-energy
photons (>1.022 MeV) interact with matter, the reverse process can also occur: the photon be
annihilated into an electron-positron pair. This process conserves charge as well as linear and
angular momentum, and is usually referred to as pair production. As with the photoelectric
eect, the annihilation of photons results in a reduction of gamma intensity.
Two other mechanisms are worth noting, though they do not change the essential physics of the
problem.
i
A fourth mechanism, Rayleigh scattering, corresponds to the elastic (photon-energy-
conserving) scattering of light, which simply changes the direction of incident photons. Though
no energy is lost, this process will reduce the measured intensity of radiation simply because some
photons will be deected out of the detection region. Essentially, at a given energy this may be
regarded as a geometrical factor.
Finally, in principle must also consider direct absorption of incident photons by the nucleus itself,
photonuclear absorption. This process usually results in the ejection of one or more neutrons and/or
protons. This interaction can contribute 510% to the total photon interaction, though within
a fairly narrow energy region usually occurring somewhere between 5 MeV and 40 MeV. At the
energies of the present experiment (<2 MeV), this eect is negligible.
6.1.1 Gamma Absorption
In the simplest model, we do not distinguish between the dierent types of interactions discussed
above, but simply presume that at a given incident energy, the number of gamma rays N
removed from the incident beam of original intensity N
o
is proportional to the number of electrons
and nuclei along the gamma rays path through the material. That is, we assume a constant at-
tenuation coecient for the beam, which may be viewed as an energy-dependent weighted sum of
separate attenuations from all three mechanisms above.
i
Of course, there are still other mechanisms, such as nuclear-resonance scattering and Delbrck scattering, but
they are negligible from our point of view.
6.1 Introduction 41
The number of electrons and nuclei encountered is itself proportional to the density of the material
, and the path length traveled in the material x. The coecient of proportionality is known as the
mass attenuation coecient , and has units of inverse distance. The loss of gamma ray intensity
can then be expressed as
N = N
o
(x)
_
_
(6.1)
Typically, one deals with the density-normalized mass attenuation coecient /, which has units
of cm
2
/g, rather than the raw mass attenuation coecient. Similarly, the quantity x is known as
the mass thickness, with units of g/cm
2
. In any case, the solution to this rst-order dierential
equation is well-known:
N(x) = N
o
e
(x)(/)
(6.2)
N(x) is the number of gamma rays remaining after passing through a material of thickness x.
Two reduced quantities determine the gamma ray attenuation: the density-normalized attenuation
coecient /, and the mass per unit thickness (often called mass thickness) x, which has units
of g/cm
2
. An exponential law is not unexpected in this case, one obtains the same solution for any
system in which the primary quantity grows or diminishes at a xed rate.
6.1.2 Attenuation Coecient
The simplistic model outlined above would certainly seem reasonable for a single energy-independent
mechanism for beam attenuation. Considering all three interactions, however, we must at least con-
sider the density-normalized attenuation coecient to be a function of energy. Typically, the quoted
coecients are a sum of attenuation coecients for the principle photon interactions:
=
pe
+
coh
+
incoh
+
pp
(6.3)
Here
pe
is the coecient for the photoelectric eect,
coh
for coherent (Rayleigh) scattering,
incoh
for incoherent (Compton) scattering, and
pp
for pair production. Overall, when dealing
with monochromatic radiation, one may simply consider an eectively constant at that particular
energy, and not worry about the detailed contributions to the eective .
However, each process has a characteristic energy dependence, making a strongly varying function
of energy. The photoelectric attenuation decreases rapidly with increasing energy, leading to an
overall decreasing background for (E). The Compton eect has a roughly constant attenuation
until 0.11 MeV, after which it decreases rapidly with increasing energy. Pair production is
possible only above 1.022 MeV, and increases rapidly for higher energies. Taken together, (E) is
a strongly decreasing function of energy until a few MeV, after which it increases gently, shown in
42 6.2 Objective
Fig. 6.2. At lower energies, one must also factor in characteristic x-ray absorptions.
ii
0.1 1 10 100
10
-9
10
-7
10
-5
10
-3
10
-1
10
1
Pb (Z=82)
Coherent scattering
Incoherent scattering
Photoelectric absorption
Pair production (nuclear field)
Pair production (e
-
field)
Total

C
r
o
s
s

s
e
c
t
i
o
n

(
c
m
2
/
g
)
Photon Energy (MeV)
0.1 1 10 100
10
-10
10
-8
10
-6
10
-4
10
-2
10
0
Al (Z=13)
Coherent scattering
Incoherent scattering
Photoelectric absorption
Pair production (nuclear field)
Pair production (e
-
field)
Total

C
r
o
s
s

s
e
c
t
i
o
n

(
c
m
2
/
g
)
Photon Energy (MeV)
Figure 6.2: (left) The absorption cross-section of lead (Z = 82) for gamma rays showing the contributions from various
processes. (right) The corresponding data for aluminum (Z =13). Data from the NIST XCOM database http://www.nist.
gov/physlab/data/xcom/index.cfm
Figure 6.3 shows the absorption cross-section for Pb as a function of photon energy. From this one
can see the superimposed eects of Compton scattering, photoelectric eects, and pair production
as well as characteristic X-ray absorption energies. At the energies used in this experiment (
57
Co
and
137
Cs gammas), the value of should vary by approximately a factor of 2030.
6.2 Objective
The primary purpose of this laboratory is to determine the mass absorption coecients for lead,
copper, and aluminum for a xed gamma ray energy (662 keV) to ascertain their utility in shield-
ing ionizing radiation. A secondary purpose will be to ascertain any energy dependence of the
attenuation factor for lead by comparing the attenuation factors at two dierent energies (662 keV
and 121 keV). Figure 6.3 shows the mass attenuation coecient for lead for energies up to 20 MeV.
The decaying background is a superposition of the eects of Compton scattering and photoelectric
eects, while the sharp increases are due to the characteristic X-ray absorptions of Pb.
Hypothesis: The attenuation of gamma rays by a solid material can be described by a sin-
gle parameter, the mass attenuation coecient. This quantity should be energy-dependent, and
material-dependent.
ii
The X-ray absorption energies correspond to incident photons matching the energy level spacing of the atoms
in the detector material. This leads to strong absorption at these particular energies, and hence, strongly increased
attenuation.
6.3 Preparatory Questions 43
10
-3
10
-2
10
-1
10
0
10
1
10
2
10
-2
10
-1
10
0
10
1
10
2
10
3
10
4
M
1-3
L
1-3
K
5
7
C
o

p
h
o
t
o
p
e
a
k
s
1
3
7
C
s

p
h
o
t
o
p
e
a
k

M
a
s
s

a
t
t
e
n
u
a
t
i
o
n

c
o
e
f
f
i
c
i
e
n
t

(
c
m
2
/
g
)
photon energy (MeV)
Pb (Z=82)
Figure 6.3: Density-normalized mass attenuation coecient / for Pb (Z =82) as a function of photon energy. Note the
logarithmic scales. The energies of primary
57
Co and
137
Cs gamma ray photons are noted, as are the K, L, and M absorption
energies for Pb.[17]
You should touch on these questions in your report.
1. Where are gamma rays on the electromagnetic spectrum?
2. Can we be completely shielded from gamma rays? If not, then why bother?
3. Which material provides the best shielding against gamma radiation? Does the answer change
as the gamma energy changes?
4. What are the possible ways in which photons with energies in the range from 1 to 2000 keV
can interact with matter?
6.3.1 Relevant Reading
Taylor[1], Ch. 1-3
Pfeer & Nir[2], Ch. 5.3.1-3, 5.5.1
http://en.wikipedia.org/wiki/Gamma_ray
See [1719] for relevant nuclear data and [20] for example gamma spectra.
6.4 Supplies
2. NaI(Tl) detector
4. lead cylinder for shielding
5. sheets of Al, Cu, and Pb
6. radioactive sources
7. detector holder and shelf housing
8. calipers for measuring sheet thickness
Pb brick
Pb cylinder
detector
detector
housing
sources
control electronics
calipers
Pb, Al, Cu
sheets
Figure 6.4: The Spectech UCS30 spectrometer and accessories.
6.5.1 Startup
2. run the program UCS30 from the shortcut on the desktop
3. place the
137
Cs source on top of the lead brick, and place the lead cylinder around it
4. place the cylindrical NaI detector inside the lead cylinder
will take several minutes. The nal spectra should look like the one shown below in Fig. 6.5.
1. Once the calibration is nished, place the NaI detector on top of the slotted stage with the
lead cylinder around it
2. Place the
137
Cs source in the third slot from the top using the plastic holder and record the
source-detector distance
0 200 400 600 800
10
100
1k
10k

137
Cs
I
n
t
e
n
s
i
t
y

(
c
o
u
n
t
s
)
E (keV)
660 keV
Photopeak
478 keV
Compton edge
184 keV
Backscatter
~30, 70-90 keV
Cs, Pb X-rays
Figure 6.5: Gamma spectrum of
137
Cs. Aside from the main photopeak at 660 keV, clear Compton edge and backscatter
features are visible, as well as X-ray absorption edges. Note the logarithmic scale on the vertical axis.[3]
3. Acquire a spectrum of the
137
Cs source alone for a preset amount of time. An example
spectrum is shown in Fig. 6.5. In the gamma ray spectroscopy unit, you will learn more
about the interpretation of these spectra.
From Settings/Presets/Time choose a live time of 300 seconds
Click the eraser icon on the toolbar to erase the existing spectrum
Click the Go button on the toolbar to begin recording your spectrum. When the
allotted time has elapsed, the Go button should be undimmed and the Stop button
dimmed. A large peak around 660 keV should be visible.
Save the spectra (name it sensibly!), and perform a data analysis from Display/Data
report record the peak centroid, gross counts, net counts, full-width at half-maximum
(fwhm).
iii
From the main screen, record the number of counts at the centroid position.
If you do not have at least 500 counts for the 660 keV photopeak, increase the live
time as required.
4. Keeping your source at the same position, insert a sheet of Pb between the source and detector.
Measure the thickness of this sheet with the calipers provided. Use gloves when handling
Pb, and wash your hands thoroughly when nished.
5. Acquire and save the whole spectrum using the steps above and record the data for the 660 keV
peak.
6. Repeat for increasing numbers of Pb sheets until you have a peak intensity at least 3 times
iii
You can also do this graphically from the main screen if you prefer.
46 6.6 Data analysis
lower than you had without any Pb or you run out of sheets.
7. Record a spectrum without any Pb sheets to verify the unattenuated intensity.
8. Repeat the steps above for sheets of Cu or Al. You may need to add 2-4 sheets of Al at a
time to achieve reasonable attenuation.
Once you have acquired all of your data for all three materials using the
137
Cs source, if time
permits insert the
57
Co source. Using the 120 keV gamma ray peak, repeat the procedure above.
An example spectra for
57
Co can be found in on page 6.8.
Note: Since you have recorded the entire
137
Cs spectrum each time, you can use several of the peaks
present (not just the 660 keV peak) to plot an energy dependence of the attenuation coecient.
6.6 Data analysis
As discussed briey above, a narrow beam of monoenergetic photons with an incident intensity
I
o
iv
, penetrating a layer of material with thickness x and density emerges with intensity I given
by an exponential attenuation law
I
I
o
= exp
_
_
(x)
_
(6.4)
We can put this in a form more amenable to analysis by taking the natural logarithm of both sides:
ln I = ln I
o
(x) (6.5)
Given a material of known density and thickness, it is clear that if we plot ln I (y-axis) versus x (x-
axis), we will obtain a straight line of slope /. Thus, by measuring the gamma ray intensity as a
function of intervening material, we can extract the density-normalized mass attenuation coecient.
For each attenuating material, make a table recording the thickness of the material and the corre-
sponding counts at the peak position. Be sure to include data for zero thickness, i.e., the detector
and source along. From this table, construct a plot with ln I on the y axis and the product of thick-
ness and density (x) on the x axis. Perform a linear regression analysis to determine the slope of
the plot, which should be /. In order to compare with the NIST database, it is convenient to
have your thickness in cm, and the material density in g/cm
3
(material densities can be found in
on page 6.8). Be sure to include proper uncertainties in your determination of / (e.g., arising
from peak counts uncertainty and thickness uncertainty). An example plot is shown in Fig. 6.6
Determine / for Cu, Pb, and Al using the 660 keV gamma ray of
137
Cs, and investigate the
energy-dependence of / for Pb using the 120 keV gamma ray peak of the
57
Co (and the 835 keV
peak of
54
Mn if you have time).
iv
Intensity is proportional to the number of photons detected, so this is not at odds with our previous discussion.
0 2 4 6 8 10 12
5.2
5.4
5.6
5.8
6.0
6.2
6.4

l
n
(
I
)
x (g/cm
2
)
Pb shielding
137
Cs photopeak
662 keV
Figure 6.6: Natural logarithm of
137
Cs gamma ray (661 keV) intensity as a function of lead mass thickness x.[3] The accepted
value of / for Pb at 661.6 keV is 0.105 cm
2
/g.[4, 17]
course of all of the one-week experiments you must do each type of report (and two formal lab
reports).
Find the accepted values of the density-normalized mass attenuation coecients for Al, Pb, and
Cu.
v
How do your values compare?
The aluminum cover on your detector is about 3.2 mm thick. What fraction of the 661.6 keV gamma
rays are absorbed when passing through the aluminum cover? How about the 122 keV gamma rays?
Which material provides the best shielding against gamma radiation? Does your answer depend
on the energy of the gamma rays? The online NIST database may be helpful.
v
See, e.g., http://physics.nist.gov/PhysRefData/XrayMassCoef/cover.html.
Compare the range for gamma ray penetration to characteristic values for beta and alpha emission.
How and why are they dierent?
Theoretical analysis of the photoeect interaction for E
m
e
c
2
suggests that
= KZ
n
E
3
or ln = ln K + nln Z 3 ln E
(6.6)
where K is a constant of proportionality, Z is the target atomic number, n is a coecient between
4 and 5, and E
is the gamma energy in keV. Compton scattering eects are neglected here, they
contribute very little to the absorption.
Plot the natural logarithm of your 662 keV mass absorption coecients as a function of the natural
logarithm of the atomic number of the absorbing material. Perform a linear regression analysis to
nd the slope and intercept of best t. Based on the equation above, the slope should yield n. Is
it between 4 and 5 as the model suggests?
For gamma energies near or larger than 511 keV the equation above is no longer valid. However
it is tempting to determine the energy dependence anyway. If you measured / for lead at more
than two energies, graph the logarithm of / as a function of the logarithm of the gamma energy.
If your data resemble a straight line, is the slope close to three?
Compare the spectrum from the
137
Cs source with no absorber to the spectrum from the
137
Cs
source with a thick lead absorber. What is the origin of the gammas between the photopeak and
the Compton edge? What happened to the Ba K X-rays? The same questions arise with the
57
Co
spectrum. What has happened?
Electromagnetic Spectrum
6.8 Discussion and topics for your report 49
Figure 6.7: The electromagnetic spectrum. From http://en.wikipedia.org/wiki/Electromagnetic_spectrum.
Density of various materials
Table 6.1: Densities of absorber materials
Element Density
g/cm
2
Al 2.70
Cu 8.96
Pb 11.35
Gamma sources available
Table 6.2: Disk gamma source set
Isotope Activity (Ci) Half-life Peaks (keV)
133
Ba 1 10.8 yrs 81, 276, 303, 356, 384
109
Cd 1 462 days 88
137
Cs 1 30.2 yrs 662
57
Co 1 272 days 122, 136
60
Co 1 5.27 yrs 1173, 1333
54
Mn 1 313 days 835
22
Na 1 2.6 yrs 511, 1275
unknown 1.5 ? ?
Sources from Spectrum Techniques, Oak Ridge, TN (www.spectrumtechniques.com). USNRC
and state license exempt quantities.
Additional data for
137
Cs and
57
Co
A
137
Cs nucleus can decay via two routes to the
137
Ba ground state, shown schematically in Fig. 6.8.
First, a single beta emission of 1.176 MeV can occur with no subsequent gamma emission, bringing
the
137
Cs directly to the ground state. However, this single-step process occurs for only about
6.5% of all nuclei. The other 93.5% decay via two-step process. First, a 514 keV beta is emitted,
resulting in metastable and short-lived
137
Ba
. The metastable
137
Ba
subsequently (2.55 min half

life) decays to the
137
Ba ground state by emission of a 662 keV gamma.
Though
137
Cs has only a single gamma emission
vi
, there are x-ray emissions that can be observed
in the energy range under study. We list a few of the higher energy emissions below.
vi
Strictly speaking, there is a 283.53 keV gamma, but its intensity is about 10
5
times smaller than the 662 keV
peak.
137
55
Cs
137
56
Ba
7/2 +
11/2
3/2 +
93.5%
6.5%
137
56
Ba
30 yr
2.55 m
stable
1176 keV
0.662 keV
0 keV
Figure 6.8: Energy level diagram showing the decay of
137
Cs to
137
Ba. The
137
Cs decays via two paths, 93.5% follow a
two-step process resulting in the emission of a 662 keV gamma ray.[4] The decay of the Ba nucleus from its excited state is an
internal transition, and contributes to the decay percentages but not the gamma decay factor.[4]
Table 6.3: X-rays from
137
Cs. From [18]
E (keV) I (%) Assignment E (keV) I (%) Assignment
31.45 0.00026 Ba K
3
36.38 0.68 Ba K
1
31.72 2.04 Ba K
2
36.65 0.0079 Ba K
5
32.19 3.76 Ba K
1
37.26 0.215 Ba K
2
36.30 0.35 Ba K
3
37.35 0.0481 Ba K
4
The
57
Co spectra and decay process is a bit more complicated, and we will leave the details as an
exercise to the reader. Figure 6.9 shows an energy level diagram for the decay of
57
Co to
57
Fe.
For all but 0.18% of nuclei, the decay proceeds by electron capture to create metastable
57
Fe
states, which decay to the ground state by gamma emission. This leads to the three characteristic
gamma peaks at 136.5, 122.1, and 14.4 keV. Additional peaks appear in the spectrum due to x-
ray emissions, sum peaks, etc. The feature of interest for this laboratory is the relatively intense
122.1 keV gamma.
Table 6.4: Gammas from
57
Co. From [18]
E (keV) I
g
(%) E (keV) I
g
(%)
14.41 9.16 352.4 0.013
122.1 85.60 366.8 0.0013
136.5 10.68 569.9 0.017
230.3 0.0004 692.0 0.16
339.5 0.014 706.4 0.025
0 keV
7/2
99.82%
57
27
Co
57
26
Fe
+
14.4 keV
136.47 keV
5/2
3/2
1/2
8.6 ns
98 ns
271.8 d
0 100 200 300 400

10
100
1k
10k
100k
1M

57
Co
I
n
t
e
n
s
i
t
y

(
c
o
u
n
t
s
)
E (keV)
Figure 6.9: Left: Energy level diagram showing the decay of
57
Co to
57
Fe. Not shown is a more complicated for 0.18% of
nuclei, which take a decay path through higher
57
Fe excited states. Associated gamma rays, after the initial
+
electron capture
are 136.5, 122.1, and 14.4 keV.[4] Right: Gamma spectrum of
57
Co. Note the logarithmic scale on the vertical axis.[3]
7
Electron Charge to Mass Ratio e/m
7.1 Introduction
In this experiment, you will investigate the motion of an electron beam under the inuence of a
magnetic eld, a simple method for measuring e/m, the charge to mass ratio of the electron.
i
The
method is similar to that used by J.J. Thompson in 1897.
7.1.1 Motion of Electrons in a Perpendicular Magnetic Field
A beam of electrons is rst created by thermionic emission from a heated metal lament, and then
accelerated through a known potential V
acc
. If the electrons are emitted from the lament with
negligible kinetic energy,
ii
, conservation of energy allows us to determine the electrons velocity:
1
2
mv
2
= eV
acc
= v =
2eV
acc
m
(7.1)
A pair of Helmholtz coils produces a uniform and measurable magnetic eld B at right angles to the
electron beam. Recall that in the presence of a perpendicular magnetic eld, an electron of charge
e will undergo uniform circular motion, as shown in Fig. 7.1, since the Lorentz force

F
B
=e v
B
is always at a right angle to the particles velocity vector.
If the electrons are in circular motion, then the magnetic force must provide the centripetal force
F
C
:
F
C
=
mv
2
r
= F
B
= evB (7.2)
Solving for the charge to mass ration e/m, we nd:
e
m
=
v
Br
(7.3)
Using our previous relationship for the electrons velocity,
e
m
=
1
Br
2eV
acc
m
=
e
m
=
2V
acc
B
2
r
2
(7.4)
i
Much of this procedure is reproduced from Pasco document 012-03471D, e/m Apparatus. This partial repro-
duction is only for use in PH255 at the University of Alabama, as dictated by the copyright notice in 012-04049J.
ii
A reasonable assumption, since even at a lament temperature of T =4000 K the thermal energy is only kBT
0.35 eV, giving an electron velocity of 310
5
m/s. This is to be compared to the electrons velocity after acceleration,
10
7
m/s.
53
54 7.1 Introduction
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
+
q
v
F r
Bin
+
q
v
F
+ q
v
F
Figure 7.1: An electron moving at velocity v in a plane perpendicular to a magnetic eld B undergoes circular motion.
Finally, if we know the radius of the Helmholtz coils and the current passing through them
iii
we
can determine the magnetic eld near the center of the coils:
B =
_
4
5
_
3/2
N
o
I
a
(7.5)
Here N is the number of turns in each coil,
o
=4 10
7
N/A
2
, I is the current through the coils,
and a the radius of the coils. Putting everything together, we can relate the e/m ratio to known
experimental quantities:
e
m
=
v
Br
=
2V
acc
_
5
4
_
3
a
2
(N
o
Ir)
2
(7.6)
A knowledge of the accelerating voltage, the construction of the Helmholtz coils and their current,
and a measurement of the radius of the electrons circular path is sucient to determine e/m.
7.1.2 The e/m Tube
Of course, electrons are quite small, and their velocities will be prodigious. How can one measure
the radius of the circular path? An ingenious method is to perform the experiment in side a vac-
uum tube back-lled with a small pressure of a noble gas in our case, 1.3 Pa of He. A partial
vacuum is necessary so that most electrons can travel without colliding with He atoms. However,
those few electrons in the beam that do collide with He atoms will excite them such that they
will subsequently radiate visible light, allowing us to view the trail of the electron beam. This is a
somewhat delicate balance: too much He and the mean distance between collisions for the electrons
is too small to be of any use, too little He and the glowing beam will barely be visible. A schematic
iii
Recall that in the Helmholtz geometry, the coil spacing is equal to the coil radius.
7.2 Objective 55
of the e/m tube is shown in Fig. 7.2
Helium-lled
Vacuum tube
Deection plates
Electron gun
cathode
heater
grid anode (+)
deection plates
e
-
Figure 7.2: Left: The e/m tube. Right: Electron gun.
The electron beam itself, as mentioned above, is created by thermionic emission from a heated
metal lament, part of a device known as an electron gun also shown in Fig. 7.2. In the electron
gun, the heater heats the cathode, which emits electrons. The electrons are accelerated by a po-
tential applied between the cathode and anode. The grid, which helps to focus the electron beam,
is held positive with respect to the cathode and negative with respect to the anode.
A rather unique feature of the Pasco e/m tube is that the socket rotates, allowing the electron beam
to be oriented at any angle from 0 to 90
with respect to the Helmholtz coils. Thus, you can rotate

the tube and examine the vector nature of the Lorentz force on moving electrons, changing the
path from a purely circular to helical, for instance. You can also use small permanent magnets in-
stead of the Helmholtz coils to further investigate the eects of magnetic elds on the electron beam.
Finally, the e/m tube also has two additional deection plates, which can be used to demonstrate
the eect of an electric eld on the electron beam. This can be used to give a conrmation of the
negative charge of the electron, and to demonstrate how an oscilloscope or cathode-ray tube works.
7.2 Objective
The objective of this experiment is to measure the charge-to-mass ratio of electrons e/m by applying
a magnetic eld perpendicular to an electron beam of known velocity. Introductory electromag-
netism tells us that the electrons velocity is related to the accelerating potential, and the magnetic
eld is determined by the geometry and current of the Helmholtz coils.
Hypothesis: From the accelerating potential, coil current and geometry, and the beam radius,
the ratio e/m can be determined.
56 7.3 Preparatory Questions
1. Justify the formula given for the magnetic eld near the center of a pair of Helmholtz coils.
2. Considering an ideal gas, to what pressure must the tube be evacuated for the mean free path
of He atoms to be comparable to the beam radius of 5 cm?
3. Compare thermal energy of the electrons assuming a lament temperature of 4000 K to the
kinetic energy of electrons accelerated through a 300 V potential.
4. Given a 300 V accelerating potential, what magnetic eld is required to produce a beam radius
of 5 cm?
Taylor[1], Ch. 1-3
Review of introductory electromagnetism.
7.4 Supplies
H
e
l
m
h
o
l
t
z

I
I adjust
Helmholtz
coils
e
-
tube
mirrored
ruler
focus
mode switch
up: e/m
V meter
deection
V accel.
lament I
- de - - acc -
Helmholtz I
- l -
Figure 7.3: The e/m setup and associated electronics. The black shroud is for more easily viewing the electron beam, and
can be removed.
1. Pasco e/m apparatus
Helmholtz coils: radius and separation a=0.15 m, N =130, B[T] =7.80 10
4
I [A]
Control panel: all connections are labeled.
Cloth hood: for easier viewing of the electron beam in a lighted room.
Mirrored scale: for measuring the beam radius without parallax error.
2. Pasco SF-9585A high-voltage power supply (050/0500 V DC, 07 V AC)
3. Frederiksen power supply (024 V AC/DC)
4. 4-pairs banana cables
5. lab notebook
7.5.1 Startup
1. Place the hood over the e/m apparatus if you will be working in a lighted room.
2. Flip the toggle switch up to the e/m MEASURE position.
3. Turn the current adjust knob for the Helmholtz coils to the OFF position.
4. Verify that your power supplies are plugged in, but turned o. Turn all knobs to their
lowest positions.
5. Connect your power supplies to the front panel of the e/m apparatus as shown in Fig. 7.4.
current adjust for
Helmholtz coils
focus
knob
mode toggle:
up = e/m
down = deection
power supply
Helmholtz coils
6-9 VDC (2A max)
+ - + -
power supply
accelerating
voltage
150-300 VDC
power supply
lament current
6 V AC or DC
do not exceed 6V!
Figure 7.4: Connections for the e/m experiment. Compare the components to the labeled photo in Fig. 7.3.
1. Turn on the supplies, and adjust them to the following levels:
Filament current / Heater (SF-9585A supply, AC volts): 6 V AC. CAUTION: Do not
at any time exceed 6 V, or you will burn out the lament and destroy the e/m tube!
Accelerating voltage / Electrodes (SF-9585A supply, 0500 V DC): 150 to 300 V DC.
The 0500 V DC output uses the central and right-most plugs of the group of three
plugs. Note the small button near the voltage/current display to toggle the reading
between 50 and 500 volt outputs.
Helmholtz coils (Frederiksen supply): 69 V DC. Turn the current limit knob to about
3/4 of its full range. Note that you are using the DC output on the left-hand side of the
supply!
2. Slowly turn the current adjust knob for the Helmholtz coils (on the e/m panel) clockwise.
Watch the ammeter on the Frederiksen supply and take care that the current does not ex-
ceed 2 A. If the current is too large, the beam will be near the edge of the tube, and your
measurements will be distorted by refraction.
3. Wait several minutes for the cathode to heat up. When it does, you will see the electron
beam emerge from the electron gun and it will be curved by the eld from the Helmholtz
coils. If it is not, gently turn the tube until it is. Do not take the tube out of its socket, as
you rotate the tube the socket will turn.
7.5.2 Measurement of e/m
In order to determine e/m accurately, you will need to make several measurements of the beam
radius for varying V
acc
. For your rst measurement, adjust the current in the Helmholtz coils,
the accelerating voltage, and tube position until you have a nice circular path of radius 5 cm.
Suggested starting parameters: I
coil
1.5 A, V
acc
250 V.
1. Adjust the focus knob as necessary to sharpen the beam.
2. Align the rotational position of the tube to obtain a closed circular path. While doing this,
qualitatively investigate the helical path of the beam when the beam is not aligned with the
coils.
3. Read the current to the Helmholtz coils from the display on the Frederiksen supply and
the accelerating voltage from the SF-9585A supply. Record these values in your laboratory
notebook, following the example of the table provided below on page 61.
4. Carefully measure the radius of the electron beam. Look through the tube at the beam. To
avoid parallax errors, move your head to align the electron beam with the reection of the
beam that you can see on the mirrored scale. Measure the radius of the beam as you see it
on both sides of the scale, and average the results. Record your values, noting your estimated
error.
5. Repeat the steps above for at least 10 accelerating voltages between 200500 V.
6. Repeat the experiment with a dierent current through the Helmholtz coils (and the same
10 accelerating voltages).
7. When nished, slowly reduce the current in the Helmholtz coils using the current adjust knob
on the e/m panel. Then reduce all voltages to zero and turn o the power supplies.
There are a few simple ways to improve your experimental results. The greatest source of error is
in the velocity of the electrons. First, the non-uniformity of the accelerating eld caused by the
hole in the anode causes the velocity of the electrons to be slightly less than their theoretical value.
Second, collisions with the helium atoms in the tube further rob the electrons of velocity. Since
e/m is proportional to 1/r
2
, and r is proportional to v, experimental values of e/m are greatly
aected by these two eects. One simple way to minimize these errors is to measure the outside of
the beam path. A second rule of thumb is to keep the accelerating voltage as high as possible, above
250 V for best results.
iv
Note that both of these sources of error will tend to make your values of
e/m higher than the accepted value.
7.5.3 Deection of electrons by electric elds
You can also use the deection plates to demonstrate how the electron beam is deected in an
electric eld, and demonstrate that electrons are negatively charged. In the presence of an electric
eld E, the path of the beam can be shown to be a parabola, governed by the equation
y =
eEx
2
2mv
2
(7.7)
where y is the linear deection achieved over a linear distance x. Beyond the deection plates, the
trajectory is a straight line, since the electrons are moving in a eld-free region of space. Clearly,
the direction of the deection will depend on the sign of the electron charge.
1. Set up the equipment as described above, except:
Perform these alterations before turning on any supplies.
Flip the toggle switch to ELECTRICAL DEFLECT.
Do not supply current to the Helmholtz coils (thus you can leave the Frederiksen supply
turned o).
Connect the DEFLECT PLATES connectors to the 050 V terminals on the SF-9585A
supply. Note that you can stack the banana plugs on the central connector, which is
common for the 050 V and 0500 V outputs.
2. Apply 6 V AC to the lament/heater and 150300 V for an accelerating voltage. Again, wait
several minutes for the cathode to warm up.
3. When the electron beam appears, slowly increase the voltage to the deection plates from 0
to approximately 50 V DC. Note the deection of the beam. The beam should be bent toward
the positively charged plate (the upper plate, if you followed the polarity conventions).
4. Warning: do not leave the beam on for long periods of time in this mode. The beam will
eventually wear through the glass walls of the tube!
7.6 Data analysis
Using the analysis above, you should be able to determine e/m for each accelerating voltage at
each current through the Helmholtz coils. Be sure to include error bars which properly take into
account all experimental errors, correctly propagated. Make a plot, with error bars, of e/m versus
accelerating voltage for each Helmholtz current, noting on the plot the position of the accepted
value. You may also plot the ratio of your determined values to the accepted values if you choose.
iv
Note that if the voltage is too high, the radius measurement is distorted by the curvature of the glass at the edge
of the tube.
You overall error should properly propagate the errors in the individual quantities that determine
e/m. Estimate the accuracy in the voltages and currents (due to either meter precision or uctua-
tions) and. the radius measurement. You may also repeat the analysis above taking into account
the electrons initial thermal energy (you may assume that only those electrons emitted directly
toward the anode end up in the beam).
Determine the earths magnetic eld at the laboratory location
v
Be sure to consider the alignment
of the tube with respect to the earths eld. Make the necessary corrections, or justify quantita-
tively neglecting them. What other environmental sources of static magnetic elds might aect
your results?
After considering all errors, report your nal average value for e/m with appropriate signicant
gures and uncertainty.
What determines the pitch of the helical path when the tube is not rotationally aligned? The radius?
Derive the formula for the eld near the center of a Helmholtz coil. What is the advantage of this
particular geometry?
How do you know that electrons are negatively charged from this experiment?
Based on the theoretical analysis and your knowledge of the apparatus, which quantity do you
expect to be the dominate source of uncertainty? How could you improve the experiment?
v
See, e.g., http://www.ngdc.noaa.gov/geomag/magfield.shtml.
Explain how a similar setup could be used to create a mass spectrometer for unknown charged
particles.
Example Data Table
Table 7.1: e/m experiment, group X, 23 Aug 2010
V
acc
(V) I
Helmholtz
(A) r (m) e/m (A-s/kg)
200 0.5 1.50 0.01
250 0.5 1.50 0.01 0.045 0.001
. . .
200 0.5 1.00 0.01
. . .
8
The Photoelectric eect
8.1 Introduction
In the late 1800s, scientists ran into considerable diculty in explaining the emission and absorp-
tion of light. Classical theory (Rayleigh-Jeans Law) predicted that the amount of light emitted
from a black body would increase dramatically as the wavelength decreased, whereas experiment
showed that it approached zero. This discrepancy became known as the ultraviolet catastrophe.
i
Experimental data for the radiation of light by a hot, glowing body showed that the maximum
intensity of emitted light also departed dramatically from the classically predicted values (Wiens
Law). In order to reconcile theory and experiment, Planck was force to develop a new model for
light called the quantum model. In this model, light is emitted in small, discrete bundles or quanta.
In this experiment, you will study the relationship between classical and quantum theories of light
emission. Analyzing the emission spectra from a mercury vapor lamp within the quantum model
will allow a determination of the ratio of Plancks constant to the electron charge, h/e. Using the
known value of the electron charge e, Plancks constant h can be determined.
8.1.1 Plancks Quantum Theory
In 1901, Planck published his law of radiation. In it he stated that that an oscillator, or any
physical system, had a discrete set of possible energy values or levels; energies between these values
never occur. Planck went on to state that the emission and absorption of radiation is associated
with transitions or jumps between two energy levels. The energy lost or gained by the oscillator is
emitted or absorbed as a quantum of radiant energy, the magnitude of which is expressed by the
equation
E = hf (8.1)
where E is the radiant energy, f the frequency of the radiation (sometimes labeled ), and h is a
fundamental constant of nature. The constant h became known as Plancks constant.
Plancks constant was found to have a signicance beyond relating the frequency and energy of
light, it became a cornerstone of the quantum mechanical view of the subatomic world. In 1918,
Planck was awarded a Nobel prize for introducing the quantum theory of light.
i
Much of this procedure is reproduced from Pasco document 012-04049J, h/e Apparatus and h/e Apparatus
Accessory Kit. This partial reproduction is only for use in PH255 at the University of Alabama, as dictated by the
copyright notice in 012-04049J.
63
64 8.1 Introduction
8.1.2 The Photoelectric Eect
In photoelectric emission, light strikes a material, causing electrons to be emitted. The classical
wave model of light predicted that as the intensity of incident light was increased, the amplitude and
thus the energy of the wave would increase. This would then cause more energetic photoelectrons
to be emitted. The new quantum model, however, predicted that higher frequency light would
produce higher energy photoelectrons, independent of intensity, while increased intensity would
only increase the number of electrons emitted (or the photoelectric current). In the early 1900s,
several investigators found that the kinetic energy of the photoelectrons was dependent on the
wavelength, or frequency, and independent of intensity, while the magnitude of the photoelectric
current, or number of electrons was dependent on the intensity as predicted by the quantum model.
Einstein applied Plancks theory and explained the photoelectric eect in terms of the quantum
model using his famous equation, for which he received the Nobel prize in 1921:
E = hf = K
max
+
o
(8.2)
where K
max
is the maximum kinetic energy of emitted photoelectrons, and
o
is the energy required
to remove them from the surface of the material (also called the work function). E is the energy
supplied by the quantum of light known as the photon.
8.1.3 h/e Experiment
A quantum of light, or photon, with energy hf is incident upon an electron in the cathode of a
vacuum tube. The electron uses a minimum
o
of its energy to escape the cathode, leaving it with
a maximum energy of K
max
in the form of kinetic energy. Normally the emitted electrons reach
the anode o the tube, and can be measured as a photoelectric current. However, by applying an
opposing potential dierence V between the anode and cathode, the electrons will be repelled from
the anode and the photoelectric current can be stooped. K
max
can be determined by measuring the
minimum reverse potential needed to stop the photoelectrons and reduce the photoelectric current
to zero, known as the stopping potential.
ii
The stopping potential V is thus the potential at
which the photoelectrons maximal kinetic energy is equal to their electrical potential energy,
K
max
= eV (8.3)
Using Einsteins equation,
hf = eV +
o
(8.4)
Solving for the stopping potential,
ii
In the present experiment we measure the stopping potential directly with an amplier circuit, rather than
monitoring the photoelectric current. See the appendices for details.
8.1 Introduction 65
V =
_
h
e
_
f

o
e
(8.5)
If we plot stopping potential (y) versus the frequency of incident light (x), the resulting graph
should be linear with slope h/e and intercept
o
/e. Coupling our experimental determination
of h/e from the slope of the V (f) graph with the accepted value for e, 1.602 10
19
C, we can
determine Plancks constant h.
Our experiment will explore the dependence of stopping potential on the frequency (wavelength)
of incident light. Incident light from an atomic vapor source will be split into discrete wavelengths
using a diraction grating and lens assembly. An individual wavelength will be chosen and aligned
incident on a photodiode, and an amplier circuit will use feedback to apply precisely the voltage
required to nullify the resulting photocurrent. This voltage is the stopping voltage, and it can be
measured on the output terminals of the amplier. The photodiode and amplier circuitry are
contained in a single assembly, henceforth known as the h/e apparatus. Below, we describe the
circuitry briey.
8.1.4 Theory of Operation
In this experiment, monochromatic light falls on a the cathode plate of a vacuum photodiode tube
that has a low work function
o
. Photoelectrons ejected from the cathode collect on the anode.
The photodiode tube and its associated electronics have a small capacitance which becomes charged
by the photoelectric current. When the potential on this capacitor reaches the stopping potential
of the photoelectrons, the current decreases to zero, and the anode-to-cathode voltage stabilizes.
This nal voltage between the anode and cathode is therefore the stopping potential of the photo-
electrons.
To let you measure the stopping potential, the anode is connected to a built-in amplier with an
ultrahigh input impedance (>10
13
), and the output from this amplier is connected to the output
jacks on the front panel of the apparatus. This high impedance, unity gain (V
out
/V
in
=1) amplier
lets you measure the stopping potential with a digital voltmeter.
Due to the ultrahigh input impedance, once the capacitor has been charged from the photodiode
current it takes a long time to discharge this potential through some leakage. Therefore, a short-
ing switch labeled PUSH TO ZERO enables you to quickly bleed o the charge. However, the
op-amp output will not stay at 0 volts after the switch is released, since the op-amp input is oating.
Due to variances in the assembly process, each apparatus has a slightly dierent capacitance.
When the zero switch is released, the internal capacitance along with the users body capacitance
66 8.2 Objective
coupled through the switch is enough to make the output voltage jump and/or oscillate. Once
photoelectrons charge the anode, the input voltage will stabilize.
8.2 Objective
The goal of this experiment is to determine the ratio of Plancks constant to the electron charge,
h/e. Using the known value of the electron charge e, Plancks constant h can be determined.
Hypotheses:
1. The stopping potential, and thus the electrons maximal kinetic energy, should increase with
the frequency of incident light.
2. The stopping potential should not depend on the intensity of incident light.
3. The stopping potential versus incident frequency should yield a linear graph of slope h/e.
1. Calculate the energies of the photons associated with the following radiation: violet light,
=413 nm; X-rays, =0.100 nm; radio waves, =10.0 m.
2. A light-emitting diode (LED) emits blue photons of wavelength 480 nm. What would be the
minimum voltage you would expect to apply to the LED before it emits light?
3. If a certain metal with a work function of
o
=2.5 eV is illuminated by monochromatic light
of wavelength 350 nm, what is the maximum kinetic energy of the electrons ejected in the
photoelectric eect?
4. What are the principle lines in the spectrum of a Hg discharge lamp?
5. What are the work functions for some common metals (list 3-4)? Why is K a common choice
for the cathode, instead of e.g., Ag? Hint: http://en.wikipedia.org/wiki/Work_function
Taylor[1], Ch. 2,3,8
Pfeer & Nir[2], Ch. 2.1-2
http://en.wikipedia.org/wiki/Photoelectric_effect
8.5 Supplies
1. Digital voltmeter or multimeter
2. h/e apparatus
8.6 Suggested Setup & Alignment procedure 67
3. h/e accessories
Lens/grating assembly
Light aperture assembly
Support base and coupling bar assembly
Light block (for rear window of source)
4. Filters for Yellow, Green, Blue, Violet, and UV Hg emission lines
5. Graded intensity lter
6. Hg and Na vapor light sources
coupling bar
light
source
lens/grating
support
base
h/e
apparatus
}
lters
multimeter
aperture
Figure 8.1: The photoelectric eect setup.
8.6 Suggested Setup & Alignment procedure
Before starting, examine Fig. 8.1 and the equipment list above to verify that all the required items
are in present. Your rst task is to make sure that your setup looks like that in the picture!
8.6.1 Setup
1. You will use the Hg light source rst, the Na source should be placed out of the way for now.
2. Insert the coupling bar assembly into the lower mounting groove of the Hg light source and
secure in place with the thumbscrew.
3. The Hg source has two openings, allowing two simultaneous experiments. We will perform
only one experiment, and for safety the rear opening (opposite side from the coupling bar)
68 8.6 Suggested Setup & Alignment procedure
should be blocked. Verify that a light block is in place. If the light block is not present, a
sheet of paper will suce.
4. Slide the light aperture assembly into the center mounting groove on the front of the light
source. Secure it in place by tightening the two thumb screws against the light source housing
(nger tight only).
5. The lens/grating assembly mounts on the support bars of the light aperture assembly. Loosen
the thumbscrew, slip it over the bars, and nger-tighten the thumbscrew to hold it securely.
The orientation is important! During the alignment procedure, you may need to turn the
lens/grating assembly around to achieve sharp lines.
6. Turn on both light sources, and allow the Hg lamp to warm up for a minimum of 10 minutes
before making any measurements. (The Na lamp takes longer to warm up, so it is good
to turn it on already at this point.) In the mean time, you may proceed with setup and
alignment. Check the alignment of the light source and aperture by looking at the light
shining on the back of the lens/grating assembly. The light should shine directly on the
center of the assembly; use the retaining screws to adjust if necessary.
7. The support base should already be attached to the h/e apparatus. If it is not:
(a) Remove the screw from the end of the support base rod. Insert the screw through the
hole in the support base plate and attache the rod to the support base plate by tightening
the phillips screw.
(b) Place the h/e apparatus onto the support base assembly.
8. Place the support base assembly over the pin on the end of the coupling bar assembly (see
Fig. 8.1).
9. Connect a digital voltmeter to the Battery Test terminals of the h/e apparatus (see Fig. 8.2).
Verify that the total voltage is 17 V. If it is not, ask the instructor for new batteries.
10. If the batteries are ok, connect the voltmeter to the OUTPUT terminals of the h/e appa-
ratus. Select a 2 or 20 V range on the meter (or use the autorange feature).
At this point, the setup is ready for operation. Note that the vapor light sources can become quite
hot! The next step is to verify that the light from the source is diracted into discrete lines and
focused on the photodiode detector.
8.6.2 Alignment
1. Set the h/e apparatus directly in front of the Hg vapor source. By sliding the lens/grating
assembly back and forth on its support rods, focus the light onto the white reective mask
of the h/e apparatus (Fig. 8.2). If you are not able to obtain a focused line, try turning the
lens/grating assembly around.
2. Roll the cylindrical light shield of the apparatus (behind the white reective mask) out of
the way to reveal the white photodiode mask inside the apparatus (Fig. 8.3). Rotate the h/e
apparatus about its support rod until the image of the aperture is centered on the window
8.6 Suggested Setup & Alignment procedure 69
reective
mask
Hg lines
amplier
discharge
voltage
output
battery
check
Figure 8.2: Projecting the Hg lines onto the reective mask. Note that the cylindrical light shield is closed here, so we do not
know if the blue Hg line falls on the photodiode detector!
in the photodiode mask. Then tighten the thumbscrew on the base support rod to hold the
apparatus in place.
3. As above, slide the lens/grating assembly back and forth to achieve the sharpest possible
image of the aperture on the window in the photodiode mask.
4. Turn the power switch for the h/e apparatus on.
5. Rotate the h/e apparatus about the pin on the coupling bar assembly until one of the colored
maxima in the rst order series shines directly on the slot in the white reective mask.
6. Rotate the h/e apparatus on its support base so that the same spectral maximum that falls
on the opening in the white reective mask also falls on the window in the photodiode mask.
iii
7. It is crucial that only one color falls on the photodiode window, there must be no overlap
from adjacent spectral maxima.
8. Replace the cylindrical light shield when you are satised.
8.6.3 Making Measurements
Once you have selected your spectral line, making a measurement of the stopping potential is
straightforward:
1. Press the push to zero button on the side panel of the h/e apparatus to discharge any
accumulated potential in the units electronics. This will assure that the apparatus records
only the potential of the light you are measuring. Note that the output voltage will drift with
the absence of light on the photodiode.
2. Read the output on your voltmeter. It is a direct measurement of the stopping potential for
the photoelectrons.
iii
The white reective mask is made of a uorescent material which allows you to see the ultraviolet line as a
blue line, and it also makes the violet line appear more blue. You see the actual colors of light if you hold a white
non-uorescent material in front of the mask (your hand will uoresce slightly . . . ).
70 8.6 Suggested Setup & Alignment procedure
window to white
photodiode mask
photodiode mask
window
Figure 8.3: Aligning the reective mask aperture and photodiode mask with the incident Hg line. Note that this picture has
been taken after rotating the cylindrical light shield to the open position. The focused blue Hg line falls on the window in the
photodiode mask, the system is properly aligned.
3. Repeat the steps above a few times. Monitor the stopping potential for a few tens of seconds
after zeroing the apparatus to note any initial transient behavior before settling on the
actual stopping potential.
For some spectral lines, or when varying the intensity of incident light, you will need to use a light
lter. The apparatus includes color several lters meant for use with the Hg vapor lamp, you can
nd their characteristics on page 74. The intensity lter has markings to indicate the percentage
of transmitted light through each region of the lter. To use the lters:
1. The yellow, green, and variable transmission lters have magnetic strips and mount on the
outside of the white reective mask of the h/e apparatus.
2. The blue, violet, and UV lters may be placed directly in front of the aperture, or axed to
the white reective mask using a magnetic mount from the optics component shelf. Do not
use tape!
3. The green and yellow lters are necessary, use them when you are investigating the yellow and
green spectral lines of the Hg lamp. These lters limit higher frequencies of light from entering
the h/e apparatus. This prevents ambient light from interfering with the lower energy yellow
and green light and masking the true results. It also blocks the higher frequency ultraviolet
light from the higher order spectra which may overlap with the lower orders of yellow and
green (though this may not be visible due to reduced intensity).
4. The blue and UV lters can be used to ensure spectral purity of the blue and UV lines if
there is a concern about overlap of the blue, violet, and UV lines. The violet lter is not
recommended, as it results in considerable attenuation of even the violet line.
5. The variable transmission lter consists of computer-generated patterns of dots and lines that
vary the intensity (not the frequency) of the incident light. The relative transmission can be
set to 100, 80, 60, 40, and 20%
8.7 Experimental Procedure 71
8.7 Experimental Procedure
According to the photon theory of light, the maximum kinetic energy K
max
of photoelectrons de-
pends only on the frequency of incident light, and is independent of intensity. Thus, the higher the
frequency of light, the greater its energy.
In contrast, the classical wave model of light predicted that K
max
would depend on light intensity.
The brighter the light, the greater its energy.
In this experiment, you will investigate both assertions. In the rst portion of the experiment, you
will select two spectral lines from a Hg source and investigate the maximum energy of photoelectrons
as a function of intensity. In the second portion of the experiment, you will select dierent spectral
lines and investigate the maximum energy of photoelectrons as a function of the frequency of light.
8.7.1 Experiment I: Wave vs. Quantum Model of Light
1. Set up the equipment as described in Sect. 8.6.
2. Adjust the h/e apparatus so that only one of the spectral colors falls upon the opening of the
mask of the photodiode. If you select the green or yellow line, place the corresponding colored
lter over the white reective mask on the h/e apparatus. You may also use the lters for
the Blue or UV lines if you choose.
3. Place the variable transmission lter in front of the white reective mask (and over the colored
lter, if one is used) so that the light passes through the section marked 100% and reaches
the photodiode. Record the voltmeter reading (stopping potential) in the table at the end of
this section. Record the approximate error in your measurement.
4. Press the instrument discharge button, release it, and observe approximately how much time
is required to return to the recorded voltage. Monitor the potential for at least a full minute
to be certain the stable stopping potential has been reached.
5. Move the variable transmission lter so that the next section is directly in front of the incoming
light. Record the new voltmeter reading, and approximate time to recharge after the discharge
button has been pressed and released.
6. Repeat the previous step until you have tested all ve sections of the lter.
7. Repeat the procedure for a second color in the spectrum.
8.7.2 Experiment II: Energy, Frequency, and Wavelength
1. Set up the equipment as described in Sect. 8.6.
2. You should be able to see ve colors in two orders of the mercury light spectrum. Use a sheet
of white paper to view the whole set of lines at once if necessary.
3. Adjust the h/e apparatus so that only one of the spectral colors from the rst order lines falls
upon the opening of the mask of the photodiode.
72 8.7 Experimental Procedure
4. For each color in the rst order, measure the stopping potential and record it in the table
below (including an estimate of the error in your measurement). If you select the green or
yellow line, place the corresponding colored lter over the white reective mask on the h/e
apparatus. You may also use the lters for the Blue or UV lines if you choose.
5. Repeat the process for the second order lines.
6. Replace the Hg source with the Na source and measure the stopping potential for the yellow
line (it should be the only line visible). The yellow line is a doublet at =589 nm.
8.7.3 Data Tables
Note 1 THz = 10
12
Hz.
Table 8.1: Stopping potential and charging time vs. intensity
Color % Transmission Stopping potential (V) Approx. charge time (s)
Color #1: 100
80
= 60
40
20
Color #2: 100
80
= 60
40
20
Table 8.2: Stopping potential vs. wavelength
First order color Wavelength (nm) Frequency (THz) Stopping potential (V)
yellow
green
blue
violet
UV
Second order color Wavelength (nm) Frequency (THz) Stopping potential (V)
yellow
green
blue
violet
UV
8.8 Analysis 73
8.8 Analysis
8.8.1 Stopping potential vs. intensity
1. Describe the eect that passing dierent amounts of the same colored light through the
variable transmission lter has on the stopping potential and thus the maximum energy of
photoelectrons, as well as the charging time after pressing the discharge button
2. Describe the eect that dierent colors of light had on the stopping potential and thus the
maximum energy of photoelectrons.
3. Defend whether this experiment supports a wave or quantum model of light based on your
results.
4. Explain why there is a slight drop in the stopping potential as the light intensity is decreased.
iv
5. Plot stopping potential versus intensity for each line, including error bars on the vertical axis.
8.8.2 Stopping potential vs. wavelength
1. Determine the wavelength and frequency of each spectral line.
2. Plot a graph of stopping potential versus frequency, including error bars on the vertical axis.
3. Determine the best-t slope and y-intercept of the graph. Interpret the results in terms of
the h/e ratio and the
o
/e ratio. Using the accepted value for e, calculate h and
o
.
4. In your discussion, report your values and discuss your results with an interpretation based
on a quantum model for light.
UV Safety
Ultraviolet radiation is emitted from low pressure mercury vapor discharges. The mercury vapor
emits ultraviolet radiation when an electrical discharge is passed through it most of the energy
iv
The impedance of the zero gain amplier is very high (10
13
) but not innite, so some charge leaks o. Thus
charging the apparatus is analogous to lling a bath tub with dierent water ow rates while the drain is partly open.
emitted is at a wavelength of 254 nm. This lies in the UV-C portion of the spectrum (180280 nm).
In the lamps used in this experiment, the 254 nm radiation will be absorbed by the outer glass tube
encasing the lamp. If the outer glass tube is cracked or broken, this radiation can cause severe skin
burn and eye inammation. Inspect the tube for cracks before performing the experiment. If the
glass bulb is broken, immediately turn the lamp o and ask the instructor to remove it to avoid
possible injury.
In uorescent lighting, the 254 nm radiation is used to excite a phosphor which coats the inside
of the glass envelope of the lamp. The phosphor will re-emit at visible wavelengths (dierent
phosphors produce dierent colors), and any UV-C which is not absorbed by the phosphor will be
absorbed by the glass wall of the lamp. However, the mercury discharge will also emit at other
wavelengths notably at 365 nm, which lies in the UV-A (315400 nm). This UV-A radiation
may not be absorbed by the phosphor, and much of it will pass out through the lamp walls into
the environment. For more information on UV radiation, see the Health Physics Society site at
http://www.hps.org/hpspublications/articles/uv.html.
Visible Spectrum
Figure 8.4: Linear representation of the spectrum of visible light. From http://en.wikipedia.org/wiki/Visible_spectrum.
Numbers are wavelengths in nanometers.
Lamp and Filter Reference Data
Table 8.3: Dominant Hg emission lines
Color f (THz) (nm)
yellow 518.7 578
green 549.0 546.1

blue 687.9 435.8
violet 740.9 404.7
ultraviolet 820.3 365.5
The yellow line is a doublet of 578 and 580 nm.

350 400 450 500 550 600
0
2k
4k
6k
8k
10k
12k
14k
16k
18k
5
7
8
n
m
5
4
6
n
m
4
3
6
n
m
4
0
5
n
m
3
6
5
n
m
blue filter
UV filter
yellow filter
green filter

I
n
t
e
n
s
i
t
y

(
a
.
u
.
)
(nm)
no filter
Figure 8.5: Intensity of the Hg high pressure lamp as a function of wavelength with and without various lters. The vertical
dashed lines are reference data, except the yellow 578 nm line which was determined experimentally with the Ocean Optics
USB-600 spectrometer. Note that the 578 nm line is actually a doublet of 578 nm and 580 nm. The intensity of the UV- and
blue-ltered spectra have been multiplied by ten times for visibility, and all spectra have been vertically separated for clarity.
9
Observation of Atomic Spectra
9.1 Introduction
When gases are placed in a tube and subjected to a high-voltage electric discharge, the electrons
in the atoms can be excited to higher energy levels within the atoms; when they return to their
original levels electromagnetic radiation is emitted. In this experiment a vapor (such as helium
or hydrogen) is placed in an electric discharge tube and a high voltage is placed across the tube.
The excited emission may look almost white or have a characteristic color depending on the vapor
inside, but it is in reality composed of a number of dierent colors or wavelengths of visible light.
You will rst use a simple prism spectroscope to separate and observe the dierent colors in the
visible spectrum qualitatively.
i
After observing the spectra with your own eyes, you will use a
USB-controlled optical spectrometer to more carefully measure the emission lines and their relative
intensities.
In the second stage of this laboratory, you will apply your knowledge of spectroscopic measure-
ments to study the thermal emission characteristics from two very important sources: the sun, and
normal household light bulbs. Comparing these spectra will help to illustrate the concepts behind
black-body radiation, as well as solar and atmospheric physics and color composition. You will
be able to estimate the temperature of the surface of the sun and the light bulb lament, and by
comparison with your atomic spectra from the rst stage, establish the likely presence of H, and
He in the sun.
9.1.1 Atomic Spectra
9.1.2 Hydrogen Spectrum: Balmer Series
original levels electromagnetic radiation is emitted. The resulting light intensity versus wavelength
for an excited gas should thus contain discrete peaks, corresponding to discrete atomic transitions
unique to individual atoms. Some of this radiation may be in a wavelength region that is visible
to the human eye. Similarly, when light passes through a gas, photons whose energies correspond
to these transitions will be absorbed, leading to a loss of intensity at the corresponding wavelengths.
i
There is also some emission in the ultraviolet region, corresponding to higher energy transitions, as well as
emission in the infrared region, but the human eye cannot see these transitions.
77
78 9.1 Introduction
In this experiment, a vapor (such as helium or hydrogen) is placed in an electric discharge tube and
a high voltage is placed across the tube. The excited emission may look almost white or have a char-
acteristic color depending on the vapor inside, but it is in reality composed of a number of dierent
colors or wavelengths of visible light. For the sake of discussion, we will focus on the hydrogen atom.
Figure 9.1: In the simplied Bohr model of the hydrogen atom, the emission lines correspond to an electron jump from a
higher energy level n to a lower n
with the emission of a photon corresponding to the energy dierence between the two levels.
For the visible Balmer series, n
=2 and n runs from 3 upward. From http://en.wikipedia.org/wiki/Hydrogen_spectrum.

A series in the emission spectrum of excited hydrogen gas is the set of spectral lines observed
when various excited states of the hydrogen atom decay into a common nal state, illustrated
schematically in Fig. 9.1. In the case of the Balmer series, which contains several wavelengths in
the visible range, this nal state is the n = 2 state of hydrogen. The quantity n is the so-called
principal quantum number of the hydrogen atom. The energy levels of the hydrogen atom are given
by the formula
E
n
=
E
i
n
2
n N
1
= 1, 2, 3, . . . E
i
= ionization energy = 13.6 eV (9.1)
Essential to the derivation of emission and absorption spectra is the postulate that a single light
quantum, more commonly called a photon, of energy E
n
=hf =hc/ is emitted when the atom
makes a downward transition from an excited state to a lower state (here h is Plancks constant, c
is the speed of light, f the frequency of emitted light, and the wavelength of emitted light). The
energy of the light emitted from a transition between two energy levels E
n
and E
n
is thus equal
to the photon energy,
9.1 Introduction 79
hf =
hc
= E
n
= E
n
E
n
= E
i
_
1
n
2

1
n
2
_
(9.2)
Hence, for the case of the Balmer series where the nal state n
=2 is the same for every line, we

obtain
hf =
hc
= E
n
= E
n
E
n
= E
i
_
1
2
2

1
n
2
_
(9.3)
Other series include the Lyman series (nal state n = 1, emission in the UV) and the Paschen
series (nal state n=3, emission in the infrared). Some of the lines from the Paschen series can be
observed with the USB-controlled spectrometer you will use. Often the spectral lines are labeled
, , , , etc., starting with the longest wavelength line.
For historical reasons, the formula for hydrogen emission lines is often given in terms of the principle
quantum numbers, the emission wavelength, and the Rydberg constant R
H
:
1
= R
H
_
1
n
2

1
n
2
_
(9.4)
You should be able to derive this equation using the equation of the energy levels in a hydrogen
atom and show how R
H
is related to E
i
and fundamental constants. You should also be able to
sketch the energy level diagram of hydrogen and draw in the transitions you observe in this experi-
ment. Your measurements of emission lines will ultimately give you an experimental value for R
H
,
which you can compare to accepted values.
9.1.3 Helium Spectrum
The emission spectrum from helium turns out to be considerably more complex than hydrogen.
This does not bode well for elements of increasing sophistication, of course. A full understanding
of the spectrum of complicated atoms cannot proceed without a detailed discussion of spin and
angular momentum, which is somewhat beyond the scope of this course. However, we can make
some general comments about the main features of the helium emission spectrum.
One can construct an energy level diagram for He similar to that of H. However, the fact that we
now have two electrons means that we must consider the interactions between them, and several
multi-electron atom features are already apparent in He. If we wish to study only lower-level exci-
tations of the He atom, we presume that one electron is in its 1s ground state, while the other is
excited into a higher energy level. When both electrons are in the 1s ground state, their spins are
antiparallel by Hunds rst rule.
ii
However, once excited, the electron in the upper level now has
two possibilities with regard to its spin: it may be excited into a state with spin antiparallel to the
ii
I.e., the term with maximum multiplicity lies lowest in energy.
80 9.1 Introduction
ground state electron (S=0, singlet-state para-helium), or it may be excited into a state parallel to
the ground state electron (S =1, triplet-state, ortho-helium. It is observed that the ortho-helium
states (parallel spins) lie lower in energy than the para-helium states, for a few reasons which are
rather too detailed to explain here. The main point is this: if in the H atom we had a single
possible transition, in the helium transition there are at least two similar transitions, since we must
consider the extra degrees of freedom provided by the spin quantum number. To this we must also
add the spin-orbit interaction, which further splits some of these lines into even more lines. Long
story short: helium is already a bit complicated.
For the purposes of this laboratory, we will simply consider para- and ortho-helium to be two dif-
ferent states of helium. Our helium discharge can be considered to be made up of a mixture of
para- and ortho-helium, and the observed spectra should contain a mixture of the para- and ortho-
spectra. A reference spectra is provided in the appendix, and an energy level diagram is depicted
below in Fig. 9.2.
In the Laboratory
JChemEd.chem.wisc.edu Vol. 75 No. 8 August 1998 Journal of Chemical Education 1015
Figure 2. Helium emission spectrum (see Table 2 for assignments
and relative intensities).
1
S
0
1
P
1
1
D
2
3
S
1
3
S
2
3
D
3,2,1
3
6
1
.
0

n
m
3
9
6
.
2

n
m
5
0
4
.
5

n
m
7
2
8
.
0

n
m
5
0
1
.
4

n
m
4
3
8
.
6

n
m
4
9
2
.
0

n
m
6
6
7
.
3

n
m
4
0
2
.
4

n
m
4
4
6
.
9

n
m
5
8
7
.
5

n
m
4
1
1
.
9

n
m
7
0
6
.
4

n
m
4
7
1
.
1

n
m
3
8
8
.
7

n
m
! !
E
n
e
r
g
y

/

e
V
Figure 4. Grotrian diagram for singlet (a) and triplet (b) helium
showing the transitions observed in the laboratory. The correspond-
ing wavelengths are given in Table 2.
m u i l e H f o s e i t i s n e t n I d n a s h t g n e l e v a W . 2 e l b a T
s e n i L n o i s s i m E
" ) r i a n i ( m n /
/ y t i s n e t n I
y r a r t i b r a
s t i n u
d e v r e s b O ( t n e m n g i s s A 2 1 ) n o i t a i v e D
0 . 1 6 3 5 ( 3 . 1 6 3
1
P
1
#2
1
S
0
) 3 . 0 4
a
7 . 8 8 3 3 ( 9 . 8 8 3
3
P
0 , 1 , 2
#2
3
S
1
) 2 . 0 4 6
b
2 . 6 9 3 4 ( 5 . 6 9 3
1
P
1
#2
1
S
0
) 3 . 0 6 2
a
4 . 2 0 4 5 ( 6 . 2 0 4
3
D
1 , 2 , 3
#2
3
P
0 , 1 , 2
) 2 . 0 2
b
9 . 1 1 4 5 ( 1 . 2 1 4
3
S
1
#2
3
P
0 , 1 , 2
) 2 . 0 8
a
6 . 8 3 4 5 ( 8 . 8 3 4
1
D
2
#2
1
P
1
) 2 . 0 2 1
a
9 . 6 4 4 4 ( 2 . 7 4 4
3
D
1 , 2 , 3
#2
3
P
0 , 1 , 2
) 3 . 0 9 1
b
1 . 1 7 4 4 ( 3 . 1 7 4
3
S
1
#2
3
P
0 , 1 , 2
) 2 . 0 4
b
0 . 2 9 4 4 ( 2 . 2 9 4
1
D
2
#2
1
P
1
) 2 . 0 5
b
4 . 1 0 5 3 ( 6 . 1 0 5
1
P
1
#2
1
S
0
) 2 . 0 5 2
b
5 . 4 0 5 4 ( 8 . 4 0 5
1
S
0
#2
1
P
1
) 3 . 0 5 1
a
5 . 7 8 5 3 ( 6 . 7 8 5
3
D
1 , 2 , 3
#2
3
P
0 , 1 , 2
)
D
3
e n i l
1 . 0 2 7
b
7 . 7 6 6 3 ( 8 . 7 6 6
1
D
2
#2
1
P
1
) 1 . 0 3 2
b
4 . 6 0 7 3 ( 6 . 6 0 7
3
S
1
#2
3
P
0 , 1 , 2
) 2 . 0 2 3
b
0 . 8 2 7 3 ( 1 . 8 2 7
1
S
0
#2
1
P
1
) 1 . 0 8 1
a
a
Bandpass = 2 nm, gain = 25.
b
Bandpass = 1 nm, gain = 25.
(H
$
), 410.0 nm (H
%
, h line), 433.6 (H
&
, G' line), 486.1 nm
(H
(
, F line), 656.3 (H
)
, C line) (see ref 12 for assignments).
Those for sodium and helium are shown in Tables 1 and 2
and in Figures 1 and 2. The intensities indicated in Tables 1
and 2 are not absolute, since the efficiencies of the optical
system and of the photomultiplier of the spectrophotometer
are wavelength dependent.
The students are asked to complete tasks described in
the following sections.
Task 1
Assignment of the lines in the obtained spectra by com-
parison with literature data and construction of the Grotrian
diagram (1214 ) showing the energy levels and the observed
transitions for hydrogen, sodium and helium (Figs. 3 and 4).
Verification of the selection rules *l = !1 and *j = 0, !1 for
hydrogen and sodium, and *l = !1 and *L = 0, !1, *S = 0,
*J = 0, !1 for helium, where l and L are the orbital angular
momentum quantum numbers, j and J are the total angular
momentum quantum numbers for atoms that can be con-
sidered as mono- (e.g., H and Na) and poly- (e.g., He) elec-
tron species, respectively, and S is the total spin angular mo-
mentum quantum number (1214).
Task 2
Calculation of the Rydberg constant,
H
, and the ion-
ization energy, IE, of hydrogen using the following equations
(1214):
" 1
"
=
H
1
n
1
2
1
n
2
2
(1)

IE =
H
hc
(2)
where " represents the wavelength, n
2
= 2 (Balmer series), n
1
is an integer larger than n
2
, h is the Planck constant, and c is
the velocity of light. A least squares fit of eq 1 to a plot of
1/" vs 1/n
1
2
, using n
1
= 310 for " = 656.3379.8 nm, leads
to
H
= (1.0984 ! 0.0024) 10
5
cm
1
. (All uncertainties
quoted throughout this paper correspond to the standard de-
viation multiplied by Students t parameter for a 95% con-
fidence level.) This result is in good agreement with the theo-
retical value
H
= 109,677.6 cm
1
(calculated from 1214):
"
=
#e
4
8$
0
2
h
3
c
(3)
where e represents the elementary charge, $
0
is the vacuum
permittivity, h is the Planck constant, c is the velocity of light,
and # is the reduced mass of the nucleus + electron system
given by
"
# =
m
N
m
e
m
N
+ m
e
(4)
where m
N
and m
e
are the masses of the nucleus and of the
electron, respectively. From the obtained Rydberg constant,
(1.0984 ! 0.0024) 10
5
cm
1
, and eq 2 it is possible to con-
clude that for hydrogen IE = 13.62 ! 0.03 eV, in good agree-
ment with the value IE = 13.59844 eV recommended in ref
15. It should be noted that for more accurate work the con-
Figure 9.2: Energy level diagram for singlet (S = 0) para-helium (left) and triplet (S = 1) ortho-helium (right). The
corresponding emission wavelengths are also given in table 9.2. From [12].
In fact, the Bohr model of the hydrogen atom can be rescued by introducing another parameter,
the so-called quantum defect correction
nl
. The defect correction essentially parameterizes the
degree of deviation from the Rydberg formula, and is related to the interaction of the electron
whose transitions produce the spectrum with the inner core electrons, interactions which are not
present for hydrogen. It can be considered approximately constant for levels of dierent n but with
9.1 Introduction 81
the same orbital momentum quantum number l (i.e., for the same series). For dierent series, it
takes on dierent values: it is highest for the S series
iii
and close to zero for the D series. Using
the defect correction, we can use the following empirical relationship for the D series:
1
= R
He
_
1
n
2
1
1
(n
2
3p
)
2
_
(9.5)
Here R
He
is the Rydberg constant for He, n
1
=3, 4, 5, n
2
=2, and
3p
is the (dimensionless) defect
correction for the para- or ortho- 3P state. Thus, a plot of 1/ versus 1/n
2
1
should give a straight
line of slope R
He
and intercept R
He
/(2
3p
)
2
, allowing a determination of both R
He
and
3p
.
Beyond identifying lines in your spectra due to transitions in para- and ortho- helium states, you
should attempt a determination of these two constants for both ortho- and para- He. Accepted
values are
3p
=0.009 and 0.063 for para- and ortho-helium, with
9.1.4 Black-body Radiation
Blackbody radiation or cavity radiation refers to an object or system which absorbs all radi-
ation incident upon it and re-radiates energy which is characteristic of this radiating system only,
not dependent upon the type of radiation which is incident upon it. The radiated energy can be
considered to be produced by standing wave or resonant modes of the cavity which is radiating.
The amount of radiation emitted in a given frequency range should be proportional to the number of
modes in that range. The best of classical physics suggested that all modes had an equal chance of
being produced, and that the number of modes went up proportional to the square of the frequency.
From the assumption that the electromagnetic modes in a cavity were quantized in energy with
the quantum energy equal to Plancks constant times the frequency, Planck derived a radiation
formula. The average energy E) per mode or quantum is the energy of the quantum times
the probability that it will be occupied (the Einstein-Bose distribution function):
E) =
hf
e
hf/k
B
T
1
(9.6)
where f is the frequency of radiation, T the temperature of the source, and k
B
is Boltzmanns
constant. This average energy times the density of such states, expressed in terms of either frequency
or wavelength gives the energy density, the Planck radiation formula in terms of wavelength or
frequency:
iii
This is spectroscopic notation, which you are not expected to be familiar with. For our purposes, it merely
characterizes a set of energy levels in Fig. 9.2. We have three types of levels: S, P, and D (not the same as the s, p,
and d used in atomic orbital notation).
82 9.1 Introduction
S
f
=
8h
c
3
f
3
e
hf/k
B
T
1
(9.7)
S
=
8hc
5
1
e
hc/k
B
T
1
(9.8)
Here c is the speed of light. The Planck radiation formula is an example of the distribution of en-
ergy according to Bose-Einstein statistics. An example of an approximately black body spectrum is
shown Fig. 9.3, the spectrum of an incandescent bulb at various applied voltages (and thus various
powers).
400 600 800 1000
0
1000
2000
3000
4000

I
n
t
e
n
s
i
t
y

(
a
.
u
.
)
(nm)
100V
80V
60V (x2)
40V (x4)
Figure 9.3: Spectrum of a 60 W soft white incandescent bulb at various ac supply voltages, exhibiting approximately black
body behavior.[3]
The above expressions are obtained by multiplying the density of states in terms of frequency or
wavelength times the photon energy times the Bose-Einstein distribution function with normaliza-
tion constant A=1. To nd the radiated power per unit area from a surface at this temperature,
multiply the energy density by c/4. The density above is for thermal equilibrium, so setting in-
ward=outward gives a factor of 1/2 for the radiated power outward. Then one must average over
all angles, which gives another factor of 1/2 for the angular dependence which is the square of the
cosine.
From our point of view, the interesting point is that the peak of a blackbody spectrum as a function
of wavelength can be uniquely related to the temperature of the source through Wiens displacement
law:
9.1 Introduction 83
max
=
b
T
(9.9)
where the constant b, known as Wiens displacement law, is equal to 2.8978 10
3
m K. Thus,
nding the peak wavelength will tell you the source temperature, and we will use this technique
to estimate the suns temperature and the temperature of the lament in an incandescent bulb.
Similar methods are used in infrared imaging to determine temperatures remotely.
9.1.5 Solar Spectrum
The spectrum of the Suns solar radiation is close to that of a black body with a temperature of
about 5800 K. About half that lies in the visible short-wave part of the electromagnetic spectrum
and the other half mostly in the near-infrared part. Some also lies in the ultraviolet part of the
spectrum. In addition to the black body spectrum, however, several other features are readily no-
ticeable, the most prominent of which are narrow bands or notches in the spectrum corresponding
to discrete absorptions. These absorptions have two origins: rst, one has absorption in the outer
layers of the sun, due to the radiation from the interior exciting atoms in the outer layers; second,
one has to consider absorption of solar radiation by our own atmosphere.
Figure 9.4: Solar radiation spectrum, showing the black body contribution and solar/atmospheric absorption. From http:
//en.wikipedia.org/wiki/Sunlight.
The prominent absorption lines in the solar spectrum are known as Fraunhofer lines, named for
the German physicist Joseph von Fraunhofer. In all, he mapped over 570 lines, and designated
the principal features with the letters A through K, and weaker lines with other letters. Modern
observations of sunlight can detect many thousands of lines. Your quick observations should yield
dozens, many of which can be correlated with your measured H, He, and Hg spectra to provide
84 9.1 Introduction
a strong indication that these elements are present in the suns outer layers. Figure 9.4 shows an
example spectrum, exhibiting black body behavior and solar/atmospheric absorption.
It was later discovered by Kirchho and Bunsen that each chemical element was associated with
a set of spectral lines, and deduced that the dark lines in the solar spectrum were caused by
absorption by those elements in the upper layers of the Sun. Some of the observed features are also
caused by absorption in oxygen molecules in the Earths atmosphere.
9.1.6 Ocean Optics USB spectrometer
The Ocean Optics Red Tide Spectrometer (Fig. 9.5) is a precongured, o-the-shelf spectrometer
where all of the optical bench options such as grating, and entrance slit size are already selected. It
has a wavelength range of 3501000 nm and uses a detector with 650 active pixels; thats 650 data
points in one full spectrum, or one data point per nanometer. Data programmed into a memory
chip on each Red Tide includes wavelength calibration coecients, linearity coecients, and the
serial number unique to each spectrometer. The spectrometer operating software simply reads these
values from the spectrometer, eliminating the calibration procedures normally associated with op-
tical spectrometers. Figure 9.5 also shows a diagram of how light moves through the optical bench
of the spectrometer, along with a brief explanation of the components.[13] Note that the optical
bench has no moving parts.
!
!""#$%&'()(
*"#+&,&+-.&/$0(
12#32&#4(
This appendix contains inIormation on spectrometer operation, speciIications, and system compatibility.
It also includes accessory connector pinout diagrams and pin-speciIic inIormation.
5/4(.6#(7*89:::(;/3<0(
Below is a diagram oI how light moves through the optical bench oI an USB4000 Spectrometer. The
optical bench has no moving parts that can wear or break; all the components are Iixed in place at the time
oI manuIacture. Items with an asterisk (*) are user-speciIied.
!"#$%%%&"'()*+,-(*(+&./*0&1,-',2(2*3&
See !"#$%%%&'()*(+,+-.&/012, on the Iollowing page Ior an explanation oI the Iunction oI each
numbered component in the USB4000 Spectrometer in this diagram.
"##$%%%%%$%%%$%"$%&%'! "#!
6
5
3
1
2
5
7
8
9
10
Figure 9.5: Left: USB-650 Red Tide spectrometer. Right: Components of a typical Ocean Optics USB-series spectrometer.
See text for explanation of numbered components. Adapted from the Ocean Optics operating instructions manual for the USB-
4000 spectrometer.Image credits: Ocean Optics.[13]
1. SMA connector: Secures the input ber to the spectrometer.
2. Slit: A dark piece of material containing a rectangular aperture, which is mounted directly
behind the SMA Connector. The size of the aperture (from 5 to 200 m) regulates the amount
of light that enters the optical bench and controls spectral resolution.
3. Filter: Restricts optical radiation to pre-determined wavelength regions.
9.2 Objective 85
4. Collimating Mirror: Focuses light entering the optical bench toward the grating. Light
enters the spectrometer, passes through the SMA Connector, Slit, and Filter, and then reects
o the Collimating Mirror onto the Grating.
5. Grating: Diracts light from the Collimating Mirror and directs the diracted light onto
the Focusing Mirror.
6. Focusing Mirror: Receives light reected from the Grating and focuses rst-order spectra
onto the detector plane.
7. Detector Collection Lens: An optional component that attaches to the Detector to in-
crease light-collection eciency. It focuses light from a tall slit onto the shorter Detector
elements.
8. Detector Collects the light received from the Focusing Mirror or L4 Detector Collection Lens
and converts the optical signal to a digital signal. Each pixel on the Detector responds to
the wavelength of light that strikes it, creating a digital response. The spectrometer then
transmits the digital signal to the SpectraSuite application.
9. OFLV Filters: OFLV Variable Longpass Order-sorting Filters block second- and third-order
light. Optional upgrade; note present in the current setup.
10. UV4 Detector Upgrade: The detectors standard window is replaced with a quartz window
to enhance spectrometer performance (< 340 nm). Optional upgrade; not present in the
current setup.
With the present conguration, the spectrometer can measure intensities within a wavelength range
of approximately 350 to 1000 nm. See the Ocean Optics manual provided for additional specica-
tions operational instructions.
The Red Tide Spectrometer connects to a computer via the USB port. When connected through a
USB 2.0 or 1.1 port, the spectrometer draws power from the host computer, eliminating the need
for an external power supply. The Red Tide is controlled by the SpectraSuite software, a com-
pletely modular, Java-based spectroscopy software platform that operates on Windows, Macintosh
and Linux operating systems. The operation of the SpectraSuite software is straightforward, and
instructions on its use are integrated into the laboratory procedure below.
9.2 Objective
In this experiment you will observe and measure the discrete wavelengths of dierent colors of light
emitted by atoms. You will rst observe light emitted from excited hydrogen atoms, and compare
that spectra to more complicated atoms, i.e., He and Hg. You will also characterize the spectra
from continuous thermal sources (approximate black bodies) to ascertain the source temperature.
From a solar spectra, you should also be able to identify solar and atmospheric absorption (giving
indications of the solar and atmospheric composition).
86 9.3 Preparatory Questions
Hypotheses: The absorption and emission of light by atoms should occur at discrete energies,
whereas the emission of light by a thermal source should be continuous. The hydrogen spectrum
can be well-characterized by the Bohr model and Rydberg formula, while more complicated atoms
require a more sophisticated model. Real thermal sources display approximate black body behavior,
accompanied by discrete emissions.
Comment on these questions in your report.
1. How is the Rydberg constant for hydrogen R
H
related to E
i
and fundamental constants?
2. Sketch an energy level diagram for hydrogen and contrast it with the diagram for helium.
3. What is the dierence between para- and ortho-helium?
4. Distinguish between continuous and line spectra.
5. Which is hotter: red-hot, or white-hot?
6. If a glowing body starts out red, then orange, then yellow, following the visible spectrum,
why dont we have green hot objects next?
Pfeer & Nir[2], 2.1.1-7, 3.2.1
http://en.wikipedia.org/wiki/Black_body
http://en.wikipedia.org/wiki/Sunlight (spectra part only)
http://en.wikipedia.org/wiki/Fraunhofer_line
http://en.wikipedia.org/wiki/Color_temperature (beginning only)
9.5 Supplies & Equipment
1. USB drive or other portable mass storage
2. H, He, Ar, Ne, and Hg vapor discharge tubes
3. Additional vapor discharge lamp of your choosing
4. Vapor lamp housing & power supply
5. Incandescent and compact uorescent bulbs
6. Bulb holder
7. Variable transformer, 0140 V
8. Ocean Optics USB-650 Red Tide spectrometer and PC
9. Diraction grating and spectroscope
9.6.1 USB-650 Setup
The USB-650 spectrometer should already be plugged in to a USB port on the adjacent PC, and
the input ber optic cable should be running through a hole in the dark box. Inside the dark box
you should also nd a discharge lamp power supply and a lab stand and clamp to keep the input
ber in place, i.e., pointed at the source. Basically, the setup should look something like Fig. 9.6
below.
discharge lamp
housing/supply
discharge
lamp input
ber
USB-650
dark box
Figure 9.6: The spectrometer setup.
(In or near the dark box should also be an incandescent bulb holder and a variable transformer, not
shown in the picture.) If elements of the setup appear to be missing or improperly set up, please
notify the instructor. A few words on the elements of the setup:
USB-650: this little box is the entire spectrometer. It should be plugged into a USB port
on the rear of the PC, and the optical input ber should run through a hole in the dark box.
Input ber: a ber optic cable used to couple incident light into the spectrometer. Basically:
point it at the source! At its heart it is a glass ber, and it is fragile.
1. Remove the plastic cover from the SMA 905 Connectors gently. Pulling the connector
away from the ber when removing the cover will permanently damage the ber.
2. Inspect the bers periodically to ensure that the bers are transmitting light. Broken
bers stop transmitting light. Visually inspect the bers for light transmission from time
to time.
3. Avoid coiling the ber too tightly. While the momentary bend radius of a ber is
typically 200 times the diameter of the ber, the maximum bend radius of a ber that you
x in place is 400 times the diameter of the ber (e.g. 16 cm for a 400 m ber). Bending
the ber past this threshold causes attenuation and can cause permanent damage to the
ber.
4. Avoid bending the ber in sharp angles.
5. Cover the SMA 905 connectors with the supplied caps when the ber is not in use.
6. Clean the ber ends periodically with lens paper and distilled water, alcohol, or acetone.
Avoid scratching the surface.
Dark box: just what it sounds like. The front of the box may be closed to create a completely
dark environment for the spectrometer to operate in, which will eliminate any ambient light
from polluting your spectra. Close the lid carefully, and be sure that the optical ber does
not get caught!
Discharge lamp: a glass tube with two electrodes, lled with gas. Handle with care, they
are fragile and will break if dropped. After several minutes of operation, they can become
quite hot as well!
Discharge lamp housing/supply: the discharge lamps operate at 5000 V dc. Never, ever
remove or insert lamps while the power is on, and be extremely careful not to touch the
housing electrodes even when the supply is o. When removing lamps that have been in use,
check that they are not too hot to handle. If you only replace tubes with the power o, and
avoid touching the lamp or surrounding region during operation, there is no danger of shock.
9.6.2 SpectraSuite Software
The SpectraSuite software will assist you in acquiring a spectrum of emitted light intensity as a
function of wavelength. To begin, start the SpectraSuite program from the icon on the desktop.
You should see a window like the one in Fig. 9.7.
For basic emission spectroscopy, for example to observe the spectrum of a hydrogen discharge
source, we only need to set up the acquisition parameters and store a reference spectrum with no
incident light. First, set the following parameters:
1. Set the integration time to 100 ms. This is the time taken by the CCD detector to collect
incoming photons. Longer integration times lead to better signal-to-noise ratios, but longer
acquisition times. If the integration time is too high the spectrum will saturate by exceeding
the maximum count of 4095. No harm is done, but to see the full trace the integration time
must be reduced, or the light source attenuated.
2. Set the Scans to Average to 1. This is the number of discrete acquisitions taken and averaged
together to yield a single resulting spectrum on the screen. The signal to noise ratio (S/N)
Figure 9.7: SpectraSuite software just after opening. The spectrometer should already be acquiring a spectrum in real time,
shown as a red line.
improves with for larger numbers of scans n, as

n. Keep in mind that the acquisition time
grows as n however.
3. Set the boxcar average to 0. This parameter sets the boxcar smoothing width. It averages
the counts of a group of adjacent detector elements. For example, a value of 3 averages the
3 points (or bins) to the right and 3 points to the left. The nal result would be a smoother
spectrum and improved S/N ratio. But if it is too high, a loss in spectrum resolution would
result. Normally the boxcar average is not required.
You may need to adjust these parameters during your actual measurements to improve S/N or
avoid saturation. Typically, you want to adjust the integration time to yield a maximum number
of counts at any wavelength to be 10003000, but less than 4000. Increase the number of averages
until you are satised with the S/N. Before making your measurements, only one more thing needs
to be done, recording a reference dark spectrum. This is simply a spectrum with no light source
present to verify the spectrum baseline.
1. Make sure the spectrometer input ber is inside the black wooden box (through the hole on
the right side). Do not bend the ber excessively, it can easily be broken!
2. Make sure all light sources are turned o, and carefully close the lid on the black wooden box.
Tighten the latches. The real-time spectrum should now be a at line near zero counts.
3. Click on the darkened light bulb to acquire and store a dark spectrum. If you wish, click on
the disk icon to save this reference spectra. Choose the le type to be Tab delimited so
that you can open it in, e.g., Excel later the binary formats will not be readable. Choose
the desired spectrum to be Dark Spectrum.
4. Click on the icon which shows a minus sign and a darkened bulb. This will automatically
display the scope minus dark spectrum, i.e., the current acquisition after subtracting o the
dark spectrum as a correction.
You are now ready to acquire emission spectra. Carefully open the dark box.
9.6.3 Measurement of the Hydrogen spectrum
When using the discharge lamps, there are three important safety points. First, do not insert or
remove bulbs while the power supply is on. Second, do not touch the bulb or power supply contacts
while the bulb is mounted, even if you think the supply is o! Finally, the bulbs can become quite
hot after operation, allow sucient time for them to cool before removing.
1. Verify that the discharge tube supply is turned o.
2. Carefully insert the H spectral tube in the discharge tube housing. The contacts are spring-
loaded, so gently press the bottom of the bulb into the socket, allowing the top of the bulb
to t.
3. Situate the discharge tube housing in the black box on the left side, such that the bulb faces
the right side. The cord should go through a notch in the left side of the box.
4. Situate the spectrometer input ber on the far right side of the box, pointing at the discharge
tube. You may need to adjust the height of the input ber. Do not bend the ber.
Do not tighten the clamp on it beyond what is necessary to keep the ber from
moving. The ber is fragile, and can be broken by either action.
5. Your setup should now look like Fig. 9.6.
6. Turn on the discharge source, you should see the hydrogen tube glowing.
You are now ready to acquire spectra. From the previous section, the spectrometer should already
be acquiring data in real-time and measuring in scope minus dark mode. If so, you should see a
few high, sharp peaks in your spectrum.
1. Adjust the integration time and the position the input ber such the peak intensity is below
4000 to avoid saturation of the detector. You may want to x the integration time at 10 or
50 ms and then slowly move the ber stand closer until you observe a good spectrum which
does not saturate the detector.
2. Close the dark box and acquire a spectrum. You may want to average several or many spectra
together for better signal to noise ratio (e.g., 510 averaged spectra).
3. Save one of these spectra by clicking on the disk icon. Choose the le type to be Tab
delimited so that you can open it in, e.g., Excel later the binary formats will not be
readable. Choose the desired spectrum to be Processed Spectrum.
At this point, you have a nice overview of the hydrogen spectrum, displaying all the primary peaks.
However, there may be (i.e., there are) smaller peaks hiding below your current resolution limit.
The primary problem is the large dierence in intensity of some of the peaks keeping the largest
peaks below overload intensity means that some of the smaller ones are barely visible.
This is not just a problem of the plot scale: while the integration time for the measurement
is sucient to collect data for the larger peaks, the peaks of much lower intensity will have an
insucient S/N for proper identication using the same integration time. The problem is a generic
one: we are trying to measure over a large dynamic range, and that usually means that the
measurement suers at one extreme or the other. One simple way to minimize this problem is
simply to take two spectra. We already have one at low integration times, suitable for measuring
the most intense peaks (at the expense of not seeing the smaller ones). Now we will just take a
second spectra with much longer integration time suitable for measuring the weaker peaks (at the
expense of cutting o the larger ones by saturating the spectrometer).
1. Adjust the integration time until one of the smaller peaks easily visible in the previous scan
now has 3500 counts.
2. The actual value of integration time will depend on how you have positioned the ber relative
to the source, but it may need to be as high as a few seconds, and probably about ten times
higher than your previous scan.
3. You should now see a few peaks in the spectra that were not visible before. You can also
adjust the number of scan averages to improve the S/N.
4. The acquisition times will now be several seconds long, and you will be sensitive to very low
light levels. Be sure that you have tightly closed the dark box, and that you are measuring
in scope minus dark.
5. When you are satised, save this zoomed spectra.
You can rst identify the major spectral peaks in the overview scan, and then use the zoom
scan to identify any smaller peaks. Keep in mind that if peaks in the zoomed spectra have a at
top, they have overload the spectrometer, and you should not use them for peak identication (use
the overview scan).
9.6.4 Helium spectra
Once you have completed the measurement for a hydrogen lamp, you should complete the same
measurements for the He and Hg sources. Turn o the discharge lamp source and carefully replace
the bulb. Do not access the bulb when it is turned on. The bulb may have become
quite hot during the experiment, be careful when removing it!
9.6.5 Heavier elemental spectra
As a point of comparison, choose another elemental gas (Ar, Ne, Hg, etc.) and record spectra.
9.6.6 Incandescent Bulb Spectrum
Now screw an incandescent 60 W bulb into the holder inside the dark box.
1. Plug the source into the variable transformer, and make sure the transformer dial is set to 0.
2. Plug the transformer into the power strip at the base of the table. By varying the output
voltage of the transformer from 0140 V, you are varying the power to the bulb and hence
the brightness.
3. Set the transformer to 120 V, it should glow brightly.
4. Start the spectrometer acquisition, you should observe a broad spectrum reminiscent of black
body radiation.
5. Adjust the integration time and the position the input ber such the peak intensity is below
4000 to avoid saturation of the detector. You may want to x the integration time at 10 or
100 ms and then slowly move the ber up and down on its stand until you observe a good
spectrum which does not saturate the detector.
6. Close the dark box and acquire a spectrum. You may want to average several or many spectra
together for better signal to noise ratio (e.g., 515 averaged spectra).
7. Without disturbing the setup, acquire spectra for several dierent bulb powers, noting the
voltage and current. You should see a much lower intensity at all wavelengths, but also a
shift of the peak in the spectrum to higher wavelengths.
9.6.7 Solar Spectrum
1. Carefully unscrew the clamp on the input ber and remove it from the dark box.
2. Place the stand on top of the dark box and replace the ber. Do not tighten the clamp
on the ber beyond a very gentle nger tight or you will damage it!
3. Point the input ber at the sky through the window. Note the rough appearance of where
you are pointing the ber (e.g., cloudy, white, gray, blue). Note the date and time (as you
would for any measurement . . . ).
4. Set the integration time for the spectrometer to 100 msec with 3 averages.
5. Start the spectrometer acquisition, you should observe a broad spectrum reminiscent of black
body radiation. You may need to rescale the plot. If the spectrum atlines on the top, you
have overloaded the detector. Halve the integration time, and repeat until the spectra does
not have a at top indicating saturation.
6. When the intensity levels are high (10003000 max) but not saturated (4000), you are
ready. Increase the number of averages to at least 10 until you are satised with the signal
to noise ratio. Remember, you are trying to tell the dierence between discrete absorptions
in your spectrum and noise!
7. Save the spectrum. Be sure to save it as a plain text le (not SpectraSuite format) so you
can import it into, e.g., Excel later.
9.7 Discussion & Analysis 93
9.6.8 Compact Fluorescent Bulb Spectrum
Also acquire a spectrum for a the compact uorescent light bulb, using only a setting of 120 V on
the variable transformer. As with the other measurements, position the input ber such that your
maximum intensity for any line is below 4000 to avoid saturating the detector.
9.7 Discussion & Analysis
Note: The spectrometer often shows weak peaks at 777, 804, and 845 nm which are not due to
the light sources, but characteristic of the spectrometer itself. Do not worry if you cannot identify
these lines.
9.7.1 Hydrogen Spectrum:
Identify all of the observed peaks and calculate the frequency and photon energy for each peak.
Assign for each peak the atomic transitions responsible for each peak (from which level to which
level). How do your values compare to reference data? Are there expected transitions you do
not observe? Make a plot of 1/ versus 1/n
2
. Its slope should be R
H
. From R
H
, calculate the
ionization energy of hydrogen and both compare with the accepted values.
9.7.2 Helium Spectrum:
Identify all of the observed peaks and calculate the frequency and photon energy for each peak.
Using the reference spectrum in the Appendix, try to assign for each peak the atomic transitions
responsible for para- and ortho-helium. Using the theory outlined in the introduction, calculate the
Rydberg constant and defect corrections for helium. Why is the helium spectrum so much more
complicated than the hydrogen spectrum?
9.7.3 Heavier element spectrum:
Can your data be explained in the same manner as the hydrogen or helium spectra? Can you
index some or all of the observed spectral lines to known atomic energy level transitions, or at least
known spectral lines? Locate a reference spectra and compare your results.
9.7.4 Solar spectrum:
Using the position of the main peak in your spectrum, and assuming the sun is a black body ra-
diator, calculate the apparent surface temperature of the sun. Does it agree with accepted values?
Away from the peak to higher (IR) and lower (UV) wavelengths, the intensity falls o more quickly
than expected for a black body radiator. Comment on possible reasons for the discrepancy (hints:
Raleigh scattering, atmospheric and window glass absorption).
Finally, in addition to the overall black-body-like behavior, you should observe a number of discrete
absorption lines and bands in the spectrum, known as Fraunhoer lines. You should be able to
locate the principle H, He, and Hg emission lines in the solar spectrum, strong evidence that these
three elements are present in the sun. Try to identify some of the other main absorptions. What
elements and molecules absorptions might you expect to nd? You should be able to discern
bands due to absorption by common atmospheric constituents (hint: try searching for Fraunhoer
lines online).
9.7.5 Incandescent Bulb Spectrum
Plot all of your spectra together, and determine the peak emission wavelength for each voltage
setting. What trend to you expect, and what do you observe? Using the peak wavelength, estimate
the bulbs temperature as a function of applied voltage and make a plot. Assuming a constant
lament resistance for the voltages used, what behavior might you expect (hint: conservation of
electrical and photon energy)?
(optional:) The integral of your spectrum gives the total incident power through the wavelength
range studied. How should this vary with bulb temperature? With applied voltage (assuming
constant resistance)? Estimate or calculate the area under each curve and plot it as a function of
voltage.
Compare your spectra qualitatively to the solar spectrum. Can you understand why incandescent
bulbs at particular powers are favored for indoor lighting? Why is color temperature used to
characterize such lighting sources?
9.7.6 Compact Fluorescent Bulb Spectrum
Can you identify any of the lines present (hint: phosphors coat the bulb glass)? Can you comment
on why some people nd uorescent lighting harsh or unnatural compared to incandescent
or solar lighting? Why does the bulb appear white, despite the presence of discrete emission
wavelengths? Does the concept of a color temperature make sense for this source?
reports).
Reference Spectral Data
Table 9.1: Hydrogen emission lines, UV-Vis-NIR[14]
(nm) I (a.u.) Assignment
383.538 5 Balmer 9 2
388.905 6 Balmer 8 2
397.007 8 Balmer 7 2
410.174 15 Balmer 6 2
434.047 30 Balmer 5 2
486.133 80 Balmer 4 2
656.272 120 Balmer 3 2
656.285 180 Balmer 3 2
954.597 5 Paschen 8 3
1004.94 7 Paschen 7 3
Spin-orbit split lines, 2p

1/2
1s and 2p
1/2
1s
Table 9.2: Selected Helium emission lines, UV-Vis-NIR[1416]
Assignment
(nm) I (a.u.) (spectroscopic) (orbital)
361.3 4 5
1
P
1
2
1
S
0
1s3p 1s2s (p)
388.9 64 3
3
P
2,1,0
2
3
S
1
1s2p 1s2s (o)
396.5 26 4
1
P
1
2
1
S
0
1s2p 1s1s (p)
402.6 2 5
3
D
3,2,1
2
3
P
2,1,0
1s3d 1s2p (o)
412.1 8 5
3
S
1
2
3
P
2,1,0
1s3s 1s2p (o)
438.8 12 5
1
D
2
2
1
P
1
1s3d 1s2p (p)
447.2 19 4
3
D
3,2,1
2
3
P
2,1,0
1s3d 1s3p (o)
471.3 4 4
3
S
1
2
3
P
2,1,0
1s3s 1s2p (o)
492.2 5 4
1
D
2
2
1
P
1
1s3d 1s2p (p)
501.6 25 3
1
P
1
2
1
S
0
1s3p 1s1s (p)
504.8 15 4
1
S
0
2
1
P
1
1s2s 1s2p (p)
587.6 72 3
3
D
3,2,1
2
3
P
2,1,0
1s3d 1s2p (o)
667.8 23 3
1
D
2
2
1
P
1
1s3d 1s2p (p)
706.6 32 3
3
S
1
2
3
P
2,1,0
1s3s 1s2p (o)
728.1 18 3
1
S
0
2
1
P
1
1s3s 1s2p (p)
(o) denotes ortho-helium, (p) denotes para-helium

Table 9.3: Selected Mercury emission lines, UV-Vis-NIR[14, 16]
365.02 2800
365.48 300
366.29 80
404.66 1800
407.78 140
433.92 250
434.75 400
435.83 4000
546.07 1100
567.59 160
576.68 240
578.97 100
579.07 280
580.38 140
614.95 1000
671.64 160
690.75 250
708.19 250
709.19 250
Table 9.4: USB-650 Specications[13]
Physical
Detector Type Sony ILX511 CCD
No. of elements 2048 pixels (650 active)
Pixel size 14 m x 200 m
Pixel well depth 62, 500 electrons
Sensitivity 75 photons/count at 400 nm
Optical Bench
Design f/4, asymmetrical crossed Czerny-Turner
Focal length 42 mm input; 68 mm output
Entrance aperture 25 m wide slit
Fiber optic connector SMA 905
Spectroscopic
Wavelength range 3501000 nm
Entrance aperture 25 m-wide slit
Optical resolution 2.0 nm FWHM
Signal-to-noise ratio 250 : 1 (at full signal)
A/D resolution 12 bit
Dark noise 3.2 RMS counts
Dynamic range 2 10
8
; 1300 : 1 for a single acquisition
Integration time 3 ms to 65 s (15 s typical max)
Stray light <0.05% at 600 nm; <0.10% at 435 nm
Corrected linearity >99.8%
Visible Spectrum
10
Speed of Light
For this experiment, we will use the Pasco manual, found online at:
http://www.pasco.com/file_downloads/product_manuals/Complete-Speed-of-Light-Apparatus-Manual-OS-9261A.pdf
99
100 Chapter 10. Speed of Light
11
Plancks Constant from LEDs
In this experiment you will determine the quantity hc/e by analyzing the current-voltage charac-
teristics and light output of light-emitting diodes (LEDs). Note: this experiment is to be performed
with the measurement of e/k, Sec. 12.
11.1 Introduction
Max Planck (1858-1947) was an early pioneer in the eld of quantum physics. Around 1900 Planck
developed the concept of energy quantization to explain the spectral distribution of blackbody
radiation.[5] This idea is fundamental to the quantum theory of modern physics. Planck received a
Nobel Prize for his work in the early development of quantum mechanics in 1918. Planck proposed
that atoms absorb and emit radiation in discrete quantities given by
E = nhf (11.1)
where n is an integer known as a quantum number, f is the frequency of vibration of the atom, and
h is now known as Plancks constant. The smallest discrete amount of energy radiated or absorbed
by a system results from a change in state whereby the quantum number n of the system changes
by one.
In 1905 Albert Einstein (1879-1955) published a paper[6] in which he used Plancks quantization of
energy principle to explain the photoelectric eect. The photoelectric eect involves the emission
of electrons from certain materials when exposed to light and could not be explained by classical
models. Einstein assumed that the electrons absorbed one quantum of electromagnetic energy at
a time and that the energy of this quantum (photon) is
E = hf =
hc
(11.2)
where f is the frequency of the light and is its wavelength. In Eindsteins model, owing to the
discrete nature of light, an electron would only be ejected if the photon energy was greater than
the energy binding the electron to the metal. Einstein received the Nobel Prize in Physics for
this work in 1921. Niels Bohr (1885-1962) used Plancks ideas on the quantization of energy as a
starting point in developing the modern theory for the hydrogen atom. Robert Millikan made the
rst measurement[7] of Plancks constant in 1912. The best current value for Plancks constant
is[8]
h = 6.62606896(33) 10
34
J s = 4.13566733(10) 10
15
eV s (11.3)
101
102 11.1 Introduction
The two digits in parentheses denote the standard uncertainty in the last two digits of the value.
In this experiment, you will use the current-voltage relationship of a set of light emitting diodes
(LEDs) to measure Plancks constant.
11.1.1 Light-Emitting Diodes (LEDs)
An LED is a semiconductor device that emits electromagnetic radiation at (typically) optical and
infrared frequencies. The device is a junction between p-type and n-type semiconductors, known as
a p-n diode, usually GaAs-, GaP- or SiC-based materials. They emit light only when an external
applied voltage is used to bias the diode above a minimum threshold value. The gain in electrical
potential energy delivered by this voltage is sucient to force electrons to ow out of the n-type
material, across the junction barrier, and into the p-type region. This threshold voltage for the
onset of current ow across the junction and the production of light is V
o
.
The emission of light occurs after electrons enter into the p-type region (and holes into the n-region).
In the p-type region, the electrons are a small minority surrounded by many holes (in this context,
essentially the anti-particles of the electrons) and they will quickly nd a hole to recombine with.
In this process, the electron relaxes from the excited state (conduction band) to the ground state
(valence band) while the hole does the opposite. The diodes are called light-emitting because the
energy given up by the electron as it relaxes is predominantly released as a photon. Above the
threshold value, the current and light output increase rapidly with the bias voltage across the diode.
electron hole
p type n type
+
-
conduction band
valence band
Fermi level
band gap
recombination
light
+
long
-
short
Figure 11.1: The electrons and holes ow over a barrier at the p-n interface when they acquire sucient potential energy
through the applied bias voltage.
The most important energy scale is set by the so-called band gap E
g
, the energy dierence be-
tween the conduction and valence bands. This band gap a property of the materials that make up
the diode, and can be tuned over a wide variety of energies from the infrared to the ultraviolet. All
electrons and holes, once excited by the applied voltage V
o
, release the same energy E
g
when they
11.1 Introduction 103
relax, and thus only photons of a specic energy are emitted.
Based only on these facts, we know enough to predict the minimum voltage required for light
emission. If only photons of a specic energy (and hence, a specic color/wavelength/frequency)
are emitted, then regardless of the mechanism of conduction in the diode, we should expect nothing
to happen until the electrical potential energy supplied to the LED is equal to the energy of the
photons emitted. In other words, we imagine that the potential energy of each charge injected
into the LED is converted completely to light energy. Right at the threshold voltage, the electrical
energy is just equal to the photon energy, which should in turn be equal to the band gap. The
quanta of energy or photon has an energy E=hf, and an electron at a potential V
o
has a potential
energy eV
o
, so the relation between the photon energy and the threshold voltage V
o
, is given by
conservation of energy:
eV
o
= E
g
= hf =
hc
(11.4)
where f and are the frequency and wavelength of the emitted photons, c is the speed of light, and
e is the electronic charge. The quantity hc/e can thus be determined by separate measurements of
the threshold voltage and the emission wavelength:
hc
e
= V
o
(11.5)
The accepted value of hc/e is 1.2398 10
6
eV-m. Since the speed of light is dened exactly
in the SI system, determining hc/e is equivalent to determining h/e (which you can compare to
your photoelectric experiment, Sec. 8). Using a known value for the electron charge (for example,
obtained from your Millikan experiment later this semester, Sec. 15), you can determine Plancks
constant h by itself.
Figure 11.2: Schematic of a 5 mm round LED, one of the most common types of LEDs. From http://commons.wikimedia.
org/wiki/File:LED,_5mm,_green_(en).svg
104 11.2 Objective
11.1.2 Bamalab boxes
In this experiment, you will be using the Bamalab software and hardware to measure the current-
voltage characteristics of several LEDs in order to determine their threshold voltages. The Bamalab
box is a computer controlled, USB-driven box that has four basic functions: (1) sourcing current,
(2) sourcing voltage, (3) measuring current, (4) measuring voltage.
Current sourcing: current must always have a closed path to ow through. When using the
current source (+I
out
, I
out
), imagine that current ows from the red positive terminal (+I
out
),
through your device, back to the back negative terminal.
Voltage sourcing: the red terminal is the positive voltage, the black is the negative voltage.
Applying a potential dierence of, e.g., 1 Volt means that V
red
V
black
=1 V.
Voltage measurement: the same polarity convention applies - red is positive, black is negative.
Measuring a potential dierence of, e.g., 1 Volt means that V
red
V
black
=1 V.
Current measurement: the same polarity convention applies - red is positive, black is negative.
Measuring current is slightly dierent, however: current must ow into the +I
in
terminal and back
out of the I
in
terminal to be measured, thus the current meter is in series with your device.
11.2 Objective
In this experiment, you will learn how to perform basic current-voltage characteristics for electrical
devices and measure the emission wavelength of a monochromatic light source.
Hypothesis: Comparing the threshold (turn-on) voltage for light-emitting diodes to their emit-
ted wavelength, one can determine hc/e.
1. A light-emitting diode (LED) emits blue photons of wavelength 480 nm. What would be the
minimum voltage you would expect to apply to the LED before it emits light?
2. The wavelength of a green light is 5.5 10
7
m. What is the approximate frequency of this
kind of light? What is the photon energy?
3. Why cant we use incandescent or uorescent lights for this experiment?
11.4 Supplies 105
Taylor[1], Ch. 2, 3,8
Pfeer & Nir[2], Ch. 2.2.1-3
11.4 Supplies
1. Bamalab electrical supply / measurement box
2. 6 banana cables
3. 4-5 LEDs (infrared, red, green, blue, UV)
5. (optional) Ocean Optics USB-600 Red Tide spectrometer or diraction grating
11.5.0.1 Starting the Bamalab Software
From the Start menu, choose All Programs and then open A Circuits Tutorial. You should get
a window like the one in Fig. 11.3 below. If you see Labjack Found on the window, everything
is functioning properly. If your window says Demo Mode on it, ask for assistance.
Figure 11.3: Introductory screen after opening the Bamalab software.
11.5.0.2 Checking your Bamalab box
Occasionally, components of your Bamalab box may stop functioning, or at least you may have
reason to believe so. Fortunately, there is a simple procedure for verifying your Bamalab box.
This procedure presumes that the previous stage - starting the software and verifying the USB
connection - has been successfully completed.
Voltage input and output First, open the multimeter panel in the software. This should
bring up a panel like the one shown in Fig. 11.4. From the Source pulldown menu, select Volt-
age, and also select Voltage from the Measure pulldown menu. After performing each action,
the middle portion of the window should change to reect your choices.
Figure 11.4: Multimeter panel.
Next, using two banana cables, connect the voltage input and output together. Do this by con-
necting +V
in
on one side of the box to +V
out
on the other side, and similarly connect V
in
to V
out
. Now click the on button under Voltage Output portion of the screen. The circle
just above the on button should now be green. Adjust the voltage output dial to some moderate
value, perhaps 1.5 or 2 V. You are now supplying voltage from the V
out
terminals.
In the Voltage measurement portion of the screen, click on. The circle above the on button
should turn red. Now you are measuring the voltage at the V
in
terminals, which are connected di-
rectly to the V
out
terminals. If everything is functioning properly, the voltage measurement should
be very close to your output voltage. Finally, change the voltage output dial to some new value
(say, 3 V). You should see the voltage measurement indicators change to follow the change in
output, in real time. Variations of 1015 % between output and input are tolerable. If the above
is successful, your voltage sourcing and measuring are OK!
Current input and output If you have just performed the previous step, turn o the voltage
output and measurement. On the multimeter screen, elect to source and measure current now. The
output and measurement regions of the screen should change after you do this. Using two banana
cables, connect the current input and output together. Do this by connecting +I
in
on one side
of the box to +I
out
on the other side, and similarly connect I
in
to I
out
. Operating the
current source also requires you to ip a small switch on the side of the box. Make sure this switch
is set to on.
Now click the on button under Current Output portion of the screen. The circle just above
the on button should now be green. Adjust the current output dial to some moderate value,
perhaps 3 or 5 mA. You are now sourcing current from the I
out
terminals. Next, in the Current
measurement portion of the screen, click on. The circle above the on button should turn
red. Now you are measuring the current passing through the I
out
terminals, which are connected
directly to the I
out
terminals. If everything is functioning properly, the current measurement should
be very close to your output current. Finally, change the current output dial to some new value
(say, 2 mA). You should see the current measurement indicators change to follow the change in
output, in real time. Again, variations of 1015 % between output and input are tolerable. If the
above is successful, your current sourcing and measuring are OK!
11.5.0.3 Performing the I(V ) sweep
From the dc Circuits menu on the Bamalab main screen, select Current vs. Voltage. That
should bring up a screen like the one in Fig. 11.5 below.
Figure 11.5: Current-voltage panel.
1. Connect the positive end of your LED (longer leg) to +V
out
. Using a second wire, connect
+V
out
and +V
in
together.
2. Using a third wire, connect the negative side of the LED to +I
in
, and with a fourth wire,
connect +I
in
to V
in
.
3. Finally, connect I
in
to V
out
. This should give you the circuit in Fig. 11.6, which applies a
voltage to LED and ammeter in series and measures the current through and voltage across
the LED. ?e) In the Current vs. Voltage panel you should already have open, sweep the
voltage from 04 Volts, in 50 steps or so. You should see a curve like that on the rst page.
4. Once you have performed this coarse I(V ) scan, clear the graph and perform a more detailed
scan starting just below the threshold voltage (where the current begins to rise rapidly) up
to about 34 V.
i
The main point is to capture the threshold behavior and a portion of the
rapid rise
5. When you have a suitable I(V ) characteristic, save the data (File/Save) to your USB memory
device, and measure the rest of your LEDs.
V
d
I
d
V
+
in V
in
V
out
V
+
out
I
+
in
I
in
LED
+ -
Figure 11.6: Circuit for measuring I(V ) for a light-emitting diode, with Bamalab box connections labeled. Shaded components
are functions performed by the Bamalab box.
11.5.0.4 Wavelength Determination
The wavelength determination can proceed in three possible ways. First, if you are pressed for
time you may use the tabulated wavelengths in the Appendix on page 110. Do not consider this
to be skirting your laboratory duties: you will perform optical spectroscopy in later experiments!
You can use the color-wavelength charts in the laboratory or in the Appendix to this section as a
rough verication that the table matches your particular LEDs. Keep in mind that you have two
experiments to perform during this laboratory period.
Second, if you have time, you may use the Ocean Optics USB-650 optical spectrometer. The
spectrometer can very quickly (i.e., in a few minutes) measure the emission spectrum (intensity
versus wavelength) for all of your LEDs. If you have already performed the atomic spectroscopy
experiment, and the apparatus is free, this is the fastest and most accurate method. There is a
very good chance that another group is already using the spectrometer, in which case you might
bargain with them for use of the spectrometer for 5 minutes or so . . .
i
If you notice the current dropping to zero above a certain voltage, you are probably exceeding the current limit
on the Bamalab box, which is about 8 mA. In this case, reduce the maximum voltage until your maximum current is
below 8 mA.
Third, if you have already performed the optics experiment, you may use a diraction grating and
spectroscope or projection screen to relatively quickly (15 min) measure the LED wavelengths.
The instructor can help you if you wish to use this method and have not already performed the optics
experiment. Again, if you have time. If you have not already performed the second experiment
(e/k), you may want to simply use the tabulated results.
11.6 Data analysis
11.6.1 Threshold voltage
An operational denition of the threshold voltage could be that value of the bias voltage when the
current reaches 0.01 mA. Extrapolate your I(V ) curves to where they cross 0.01 mA current and
use that as the working value of V
o
.
Another way to determine the threshold voltage is to plot log(I) (y) versus V (x), which should
give roughly linear behavior near the threshold. If you t the straight-line behavior just above
threshold to a linear trendline, you can use the resulting best-t line equation to determine the
threshold voltage, either as dened above, or by nding where the line intercepts the x axis.
Finally, you can just plot I(V ) and perform a linear extrapolation of the curve (i.e., do a linear
t and nd the x intercept). Over a range of 0.05 = 0.1 V above the perceived threshold, the
characteristic is quite linear for most LEDs, and this procedure can give quite accurate results. The
main point is that you nd a reasonable and systematic way to determine the threshold voltage,
and use the same procedure in analyzing all of your data.
11.6.2 Determination of h/e
From the discussion above, you should be able to determine h/e from your data (given that c is a
constant). Do this for all of your LEDs, taking care to include estimated uncertainties in threshold
voltage and LED wavelengths.
ii
. Find h/e for each LED. Using the accepted value for e, determine
h for each LED and the average result for all of your LEDs together (making sure to propagate the
uncertainty correctly). Since you have 3 or 4 measurements of Plancks constant, you may wish to
nd the mean and standard deviation of the mean to quote the uncertainty (see the introduction
to the counting statistics experiment, Sec. 5.).
In addition to the data analysis outline above, below are some questions and topics you may wish
to address in your report.
ii
For wavelength uncertainty, you may use the full width at half maximum of the quoted wavelengths in the
Appendix on page 110.
1. Is your value for h reasonable, i.e., does the range of the value plus or minus the uncertainty
coincide with the accepted value? If not, why?
2. What is the product of the threshold voltage and the emission wavelength? What should it
be?
3. Why is it problematic to determine the threshold voltage from the I(V ) equation governing
a diode? Hint: read the introduction to the ek lab, Sec. 12.
4. How could you make your threshold determination more accurate? Presume you have use of
an optical spectrometer or light meter.
5. How is this experiment related to the photoelectric experiment you have/will perform?
6. What materials might be used to give the colors for your particular LEDs?
7. How can you create interior white lighting with discrete LED sources?
Visible Spectrum
LED Wavelengths
The table below lists the known emission wavelengths for the LEDs provided in the laboratory, as
measured by the same USB-650 Ocean Optics spectrometer you will use in later experiments. If
you do not have sucient time to verify these measurements yourself with the spectrometer or a
diraction grating, at least make a rough verication with the color-wavelength charts in the lab
or the one above. A printed and veried color chart in the laboratory is preferred, owing to the
wide variety in the quality of color reproduction with monitors and printers.
Table 11.1: LED Wavelengths
Type Peak (nm) FWHM
(nm)
Infrared 931.7 40.7
Green 514.8 37.0
Amber 590.4 17.5
Ultraviolet 413.4 18.6
Wavelengths should be veried for your particular LEDs.
FWHM = full width at half maximum

12
Measurement of e/k
In this experiment, you will determine the quantity e/k by analyzing the current-voltage charac-
teristics of transistors. Note: this experiment is to be performed with the measurement of Plancks
constant with LEDs, Sec. 11.
12.1 Introduction
In this laboratory course, you will measure a variety of fundamental constants or combinations
of fundamental constants. In many cases, the approach seems to follow straightforwardly from
material covered in your lecture courses: the ratio of the electrons charge to mass e/m can be
measured by sending an electron beam through a region of crossed electric and magnetic elds, the
ratio of Plancks constant to the electron charge can be measured by the photoelectric eect, and
the electron charge can be measured by Millikans apparatus. It is less common in introductory
physics that we encounter Boltzmanns constant k
B
, however, and even more rare to encounter an
experiment for reliably measuring it. In this laboratory, you will measure the ratio of the electron
charge to Boltzmanns constant e/k by analyzing the current-voltage characteristics of a transistor.
In later labs, a separate measurement of the electron charge e by Millikans method will allow a
determination of k
B
.
12.1.1 p-n diodes
The deduction of k
B
essentially relies on the exponential current-voltage characteristic of a semi-
conductor p-n junction. A p-n junction is a region in a semiconductor where the dominant current
carrier changes from holes (p-type, few free electrons) to electrons (n-type, many free electrons).
This region can be created by spatially varying the doping of the semiconductor, or simply joining
two dierent semiconductors. As a result of the change in carrier type, the potential energy of an
electron changes across the boundary of the two regions, the junction. The change in potential
eectively appears as an extra contact potential dierence V
o
between the n- and p-type regions.
This means that an electron encounters a potential step of height eV
o
when moving from the n-type
to the p-type region, shown in Fig. 12.1.
For an electron to move from the n- to the p-type region, it must have an energy greater than eV
o
.
Out of the total number of electrons in the n-type region, the fraction with an energy greater than
eV
o
is given by the Boltzmann distribution, exp (eV
o
/k
B
T). These electrons with energy greater
111
energy
position
eVo
p-type n-type
e(Vo - V)
p-type n-type
zero
bias
forward
bias V
eV
Figure 12.1: Potential energy of an electron for a p-n junction with zero bias (upper) and a forward bias of V (lower).
than eV
o
will diuse into the p-type region (while those of lower energy are conned to the n-type
region), giving rise to an electron current proportional to the Boltzmann factor:
I
o
= c exp (eV
o
/k
B
T) (12.1)
Here c is a constant of proportionality. In addition, there will also be an electron current in the
opposite direction, from the p- to the n-type region. One source of the electrons making up this
current is the thermal creation of electron-hole pairs. If free electrons are present in the p-type
region of the junction, the electric eld caused by the contact potential dierence will drive them
across the junction to the n-type region. In equilibrium (no external voltage applied to the junc-
tion), there can be no net current, so this reverse current must be equal to the diusion current.
The situation becomes more interesting if we apply a forward bias voltage V . A forward
bias means that the potential is reduced in the p-type region, while reverse bias means that
the potential is raised in the p-type region. WIth a forward bias of V , the potential step at
the junction boundary is now reduced to V
o
V , and the diusion current is proportional to
exp [e (V
o
V ) /k
B
T]. The reverse current is not aected by this change, since it involves diusion
in the direction of the potential gradient rather than against it, and thus the total current is
I = c exp [e (V
o
V ) /k
B
T] c exp [eV
o
/k
B
t] = I
o
[exp (eV/k
B
T) 1] (ideal p-n diode)
(12.2)
The forward current of a p-n junction thus increases exponentially with bias voltage. Note that for
reverse bias (negative V in the equation above), the potential step at the p-n interface is increased
rather than decreased, and the current is strongly suppressed. What this means is that the p-n
junction conducts current well in one direction, and badly in the other: it is a diode! For simplicity,
we will just say diode when we mean p-n junction from now on. A schematic of the operation
of a diode is shown in Fig. 12.2
Figure 12.2: A p-n junction in equilibrium with no applied voltage. From http://en.wikipedia.org/wiki/P-n_junction.
It would seem that this relationship allows a measurement of e/k: plotting the logarithm of the
current versus voltage should yield a straight line of slope e/k
B
T. Sadly, this is not the case for a
real diode junction. As it turns out, there are other factors that control the current and the nature
of the potential step, which in the end can be modeled by a simple modication of the ideal diode
equation:
I = I
o
[exp (eV/nk
B
T) 1] (real p-n diode) (12.3)
The extra factor n is the ideality factor, and can be anywhere from 1 to 2.5 depending on the
diode and voltage range. Unfortunately, the ideality factor precludes a determination of e/k
B
from
a simple diode.
12.1.2 Transistors
The simplest realization of a transistor is simply two p-n junctions in series, making a pnp or npn
structure called a junction transistor.
i
We will not dwell on their operation any further than this
fact for the moment. A junction transistor has three characteristic regions, known as the emitter
(E), base (B), and collector (C). In an npn junction transistor, the base is a p-type region while
the emitter and collector are n-type regions as shown in Fig. 12.3.
In an npn transistor, the base-emitter (B-E) junction is a p-n junction, and the current between
the collector and base I
CB
as a function of the voltage across the B-E junction V
BE
is found to
i
See http://en.wikipedia.org/wiki/Bipolar_junction_transistor for more detail.
114 12.2 Objective
Figure 12.3: An npn junction transistor with a forward-biased E-B junction and a reversed-biased B-C junction. From
http://en.wikipedia.org/wiki/Bipolar_junction_transistor.
follow Eq. 12.2 exactly. The factors giving rise to diode non-ideality are still present in the junction
transistor, but essentially we can avoid the non-ideality by using two junctions.
ii
The non-ideal
eects which give the factor of n in Eq. 12.3 are still present and do aect the emitter current,
but these currents ow out through the base of the transistor and will not be measured. For our
determination of e/k, we will sweep the collector-base current I
CB
and measure the resulting base-
emitter voltage V
BE
.
For the range of V
BE
used in this experiment (0.11 V), the thermal energy at room temperature
is negligible compared to the electrons potential energy, eV k
B
T. Equation 12.2 applied to the
collector-base current can then be well-approximated by the ideal diode equation:
I
CB
= I
o
exp (eV
BE
/k
B
T) (12.4)
A logarithmic plot of I
CB
versus V
BE
thus has a slope of e/k
B
T:
ln I
CB
= ln I
o
+ eV
BE
/k
B
T (12.5)
From the known ambient temperature, an accurate determination of e/k
B
T can now be made.
12.2 Objective
The goal of this experiment is to determine the ratio of the electron charge to Boltzmanns constant,
e/k
B
. Using the known value of the electron charge e, Boltzmanns constant k
B
can be determined.
Hypothesis:
ii
For more detail, see [9].
12.3 Preparatory Questions 115
The collector-base current of a junction transistor is a simple exponential function of the base-
emitter voltage, whose characteristic energy is determined by e/k
B
T. The I
CB
V
BE
characteristics
of an npn junction transistor can thus be used to determine e/k
B
.
What is the signicance of I
o
?
What will the eect of measurement temperature be on your experiment?
What would the eect of Joule heating on your experiment be? How does this aect the
choice of transistor?
What is the accepted value of e/k?
Taylor[1], Ch. 8-9
Pfeer & Nir[2], Ch. 6.4.7-6.4.9
See also the papers by Inman[10, 11] and Evans[9] for more details on the procedure and analysis.
12.5 Supplies
1. 2N1724 NPN Silicon power transistor
2. ECG-130 NPN Silicon power amp
3. Keithley 220 current source
4. Hewlett-Packard 3457A multimeter
5. 4-banana cables
6. 3-aligator clips
12.6.1 Startup
For this experiment, you will not need all of the electronics present at the PH255 transport station,
so dont be intimidated by the number of devices you nd! First, you will need to power up and
verify the connections and functionality of the required instruments. Figure 12.4 is provided to
help you keep the various instruments straight. Most of the instrument settings will be software-
controlled, so you need not worry greatly about the wide variety of buttons on the instruments.
Magnet supply
sample voltmeter
sample
current source
Hall voltmeter
Hall
current source
DAC
Figure 12.4: The PH255 transport measurements setup. Not all the electronics pictured are used for this experiment.
1. Turn on the Keithley 220 current source on the right side of the apparatus (labeled sample
current source in Fig. 12.4) using the small red button on the lower left portion of the front
panel. This is the current source which will supply I
CB
. After an initialization routine, the
display should read +.0000 9.
2. The output of the current source is in the rear of the instrument, where you should nd an
adaptor and two banana cables coming out of the output triaxial connector. These two
banana cables will supply I
CB
.
3. Make sure that the Output indicator on the current source is not lit. If it is, press the
Operate button to turn o the output.
4. Turn on the Hewlett-Packard 3457A multimeter directly below the current source (small grey
button on the lower left portion of the front panel). This is the voltmeter which will measure
V
BE
. After initialization, the display should have a reading ending in VDC.
5. The input of the multimeter is on the front panel of the instrument, you will use the Input
(2 wire) terminals. Two banana cables should already be plugged in to these terminals. The
multimeter can measure a variety of parameters, you will use only the dc Volts measurement
which will be enabled in software.
12.6.2 Setting up the Circuit
The circuit and parts required are shown below in Fig. 12.5. A 2N1724 npn power transistor
should already be present at your station, ask the instructor if you do not nd it. The E, B, and
C terminals are labeled on the transistor (see Fig. 12.5). Study the circuit below before connecting
anything. Identify which instruments and which wires should connect to which terminals of the
transistor before proceeding.
IBC VBE
HP3457A K220
B
C E
2N1724
+ +
- -
Figure 12.5: Left: The circuit for measuring e/k: measuring the base-emitter voltage while sweeping the base-collector
current. Right: The 2N1724 transistor, banana plugs, and alligator clips.
1. Connect the current source to the C and B terminals of the transistor using banana cables and
alligator clips. Be careful that the clips do not touch each other or the transistor housing
iii
, and
verify that the output indicator on the current source is not lit before making a connection.
Note that B should be positive and C negative.
2. Connect the multimeter to the B and E terminals of the transistor using banana cables. For
the B terminal, you may plug the multimeter banana cable into the banana cable from the
current source you just connected. For the E terminal, use a third alligator clip with the
banana cable.
3. Verify that the alligator clips are not touching each other, transistor housing, or anything
other than the transistor terminals.
4. Verify your connections. Polarity is important, diodes only conduct easily for one direction
of current ow!
12.6.3 Setting up the V (I) sweep
From this point onward, the experiment will be software-controlled, using the PC next to the
transport setup. You will sweep the current sourced between collector and base and measure the
resulting base-emitter voltage.
iii
Actually, you may have noticed that the C terminal is connected to the transistor housing. Just dont let any
other connections touch the housing.
1. Open the program PH255 Transport on the desktop.
2. Under the Measurement menu, select V(I) to open the measurement panel.
3. On the measurement panel, you can set the current sweep parameters, run the measurement,
and save your data. Below is a screenshot of the measurement panel, we will now discuss
some of the parameters relevant to the measurement.
Figure 12.6: The voltage-current V (I) measurement panel.
Voltmeter range: the maximum full-scale voltage reading. It is easiest to use the Auto
range, which will automatically adjust the voltage measurement range as needed.
Start Delay: once the initial current is set, the program will wait for this amount of time
before proceeding with the current sweep.
Point Delay: at every current value, Averages voltage measurements will be made, with a
delay of this time between individual measurements. Increasing the point delay will increase
the measurement time, but may improve stability.
Averages: at every current value, this many voltage measurements will be made and averaged
together. Increasing the number of averages decreases the signal to noise ratio, but increases
the measurement time.
Info: this text eld will let you know what the program is currently doing.
Sample info: anything typed in this box will be saved in a log le (see below) to help you
keep track of dierent measurements.
Start I: current at which to start the sweep.
End I: current at which to end the sweep (within 10%).
Step: the increment value for the current sweep. In the normal sweep mode, the current
sweep will start at Start I and increase the current by this value until End I is reached.
In Log sweep mode, the current will be multiplied by this factor at each step in the sweep
until End I is reached. For example, if you start at 1.00E 6 A with a step of 2.0, the
current will increase to 2.00E 6 A, 4.00E 6 A, etc., until End I is reached.
Back & forth: if selected, the sweep will run from start to end, and then back to the starting
value.
Log sweep: if selected, rather than increasing the current by a constant amount at each step,
the current will be multiplied by a constant amount after each step.
Sweeps: if Back & forth is selected, this will sweep the current back and forth this many
times.
Start: start the measurement
Halt: during the measurement, you may stop at any time if you notice a problem.
Clear: clear the existing plot and data. Any unsaved data will be lost!
File menu: save the V (I) data to a plain text le for further analysis. Any data still on screen
will be saved!
The following parameters are suggested values you can start out with, though you will need to
experiment a bit to optimize your curve.
Voltmeter range: Auto
Start Delay: 1 s
Point Delay: 300500 ms
Averages: 1-3
Sample info: identifying information and conditions for your measurement
Start I: 1E 7 A
End I: 1E 2 A
Log sweep: selected
Step: 2 times increase at rst, 1.25-1.5 times for the nal measurement
Back & forth: do not select
Sweeps: 1
Once you have nished your measurement, save it in a sensible location (with a sensible name)
from the File menu. This will generate two les. If your base le name is 2N1724, then the
les 2N1724.dat and 2N1724.dat.log will be created. The former is the measurement data,
in tab-delimited ASCII format (easily imported into Excel, for example). The latter is a log le,
recording the measurement parameters, instruments used, etc., as well as any text you have typed
into the sample info box. The log le contains essentially all information required to repeat the
measurement, and can be a crucial resource in when analyzing your data. (That does not mean
you shouldnt write down all important information in your lab notebook!)
Once your setup is ready, perform several (5) measurements of V (I), varying the sweep param-
eters to obtain a good characteristic over at least 45 orders of magnitude in current. Save each
measurement in a separate le so you can make several determinations of e/k.
It is probably a good idea to rst perform relatively coarse scans (e.g., log sweep with 2 times
increase at rst) to be sure that things are operating correctly, and then increase the resolution
(e.g., 1.25-1.5 times increase) when you are sure that everything is correct.
Note: After obtaining the rst good sweep, it is a good idea to look at the data in Excel to verify
the linear behavior of ln I
BC
versus V
BE
(see below) before you make subsequent measurements. At
very low currents (10
9
A) or very high currents (10
2
A) you may see deviations from linear
behavior. In the former case, the reasons are somewhat complex, in the latter case it is likely due
to Joule heating. If you observe either behavior, adjust the range of your current scan to avoid the
non-linear regimes.
Once you have several scans, try repeating the experiment with a second transistor. If the basic
theory outlined above is correct, the results should be comparable!
12.7 Data analysis
Make a plot of ln I
BC
versus V
BE
for each transistor, it should be linear. Perform a least-squares
regression (linear trendline in Excel). From the discussion above, the slope should yield e/k
B
T for
each measurement. Estimate the temperature of the room (and likely uncertainty), and determine
e/k
B
, averaging the results for all of your measurements. Since you have several measurements of
e/k
B
, you may wish to nd the mean and standard deviation of the mean to quote the uncertainty
correctly (see the introduction to the counting statistics experiment, Sec. 5.).
Using the accepted value for e, make a determination of k
B
and your uncertainty. (In a later
laboratory, you will measure e directly.)
1. Is your value for e/k reasonable, i.e., does the range of the value plus or minus the uncertainty
coincide with e/k? If not, why?
2. What are the predominant sources of error in your experiment? Why was the 2N1724 chosen?
3. How else might you determine e/k? Are there any other experiments in this course sensitive
to the value of k?
4. Why are we sweeping current and not voltage? Why did we not use the Bamalab boxes for
this experiment (unlike the Plancks constant determination)?
5. How could you improve the accuracy of this experiment. Hint: T is a variable . . .
reports).
Simplied 2N1724 Datasheet
2N1724 NPN Silicon High Power Transistor, specications from www.datasheetcatalog.org/
datasheet/microsemi/2n1722.pdf
Table 12.1: Maximum Ratings
Ratings Symbol Value Units
Collector-Emitter Voltage V
CEO
80 Volts dc
Collector-Base Voltage V
CBO
175 Volts dc
Emitter-Base Voltage V
EBO
10 Volts dc
Max. Collector Current I
C
5.0 Amps dc
Total Power Dissipation (25
) P
T
3.0 W
Simplied ECG-130 Datasheet
ECG-130 NPN Silicon power amp, specications from the original packaging.
Table 12.2: Maximum Ratings
Ratings Symbol Value Units
Collector-Emitter Voltage V
CEO
60 Volts dc
Collector-Base Voltage V
CBO
100 Volts dc
Emitter-Base Voltage V
EBO
7 Volts dc
Max. Collector Current I
C
15 Amps dc
Total Power Dissipation (25
) P
T
115 W
13
Polarization and Diraction of Light
13.1 Introduction
In the rst portion of this experiment, Malus Law of Polarization is veried by showing that the
intensity of light passed through 2 polarizers depends on the square of the cosine of the angle be-
tween the 2 polarization axes. Adding a third polarizer makes clear the vector nature of polarization.
In the second portion of this experiment, a diraction grating consisting of a series of closely-spaced
vertical lines is used to measure the wavelength of the laser light used in the rst experiment.
13.1.1 Polarization
In the rst experiment, Laser light ( =650 nm) is passed through two polarizers. As the second
polarizer (the analyzer) is rotated, the relative light intensity is recorded as a function of the angle
between the axes of polarization of the two polarizers. The plot of light intensity, as measured
by a calibrated photocell, versus the relative angle between the polarizers will be compared with
theoretical results.
A polarizer only allows light which is vibrating in a particular plane to pass through it. This
plane forms the axis of polarization. Unpolarized light vibrates in all planes perpendicular to the
direction of propagation. If unpolarized light is incident upon an ideal polarizer, only half of the
light intensity will be transmitted through the polarizer.
The transmitted light is polarized in one plane. If this polarized light is incident upon a second
polarizer, the axis of which is oriented such that it is perpendicular to the plane of polarization of
the incident light, no light will be transmitted through the second polarizer, as depicted in Fig. 13.1.
However, if the second polarizer is oriented at an angle not perpendicular to the axis of the rst
polarizer, there will be some component of the electric eld of the polarized light that lies in the
same direction as the axis of the second polarizer, and thus some light will be transmitted through
the second polarizer.
If electric eld after passing through the rst polarizer at angle is

E
o
, then the amplitude passing
through the second polarizer will be the projection onto the angle of the second polarizer. That
is, only the component of

E
o
parallel to the second polarizer set at angle will pass, giving an
electric eld amplitude E
o
cos ( ) after passing through both polarizers. Since the intensity of
123
polarizer 1 polarizer 2
unpolarized
light
polarized
light
nothing
90
component
parallel to
polarizer 2

polarized
light
Icos
2
( -)
Figure 13.1: Light transmitted through two polarizers. If the polarizers are aligned 90
oset, essentially no light passes

through both polarizers. If the polarizers have a relative angle , the intensity after both polarizers vary as the cosine of the
relative angle squared.
light varies as the square of the electric eld amplitude, the light intensity after passing through
both polarizers is given by
I = I
o
cos
2
( ) (13.1)
This is known as Malus law, and we see that the transmitted intensity will vary from zero when
the polarizers have a relative angle of 90
to I
o
when the polarizers are aligned at 0 or 180
.
i
In
retrospect, we can use Malus law in this form to nd the transmission through a single polarizer.
A beam of unpolarized light can be thought of as containing a uniform mixture of linear polariza-
tion at all possible angles. Since the average value of cos
2
is 1/2, the transmitted intensity for
unpolarized light through a single polarizer is 1/2.
Real polarizers do not perfectly block light with polarization orthogonal to their polarization axis.
The ratio of the transmission of the unwanted (90
) component to the desired (0
) component is
called the extinction ratio, and varies from around 1 : 500 for Polaroid sheets (which we will use)
to about 1 : 10
6
for GlanTaylor prism polarizers. In practice, some light is also absorbed by the
polarizer, and the actual transmission will be somewhat lower. Simple Polariod-type polarizers
may have a transmission of 38%, whereas some birefringent prisms can be considerably higher,
nearly reaching the ideal limit of 1/2. Thus, even for crossed polarizers with a relative angle of
90
, even though the theoretical transmission is exactly zero some light will be transmitted.
i
When the polarizers have a relative angle of 90
, they are said to be crossed.

polarizer 1 polarizer 3
unpolarized
light
component
parallel to
polarizer 2
90
component
parallel to
polarizer 3
0
polarizer 2
Figure 13.2: Light transmitted through three polarizers.
In the next phase of this experiment, you will explore a curious phenomena when three polarizers
are placed in series, depicted in Fig. 13.2. The rst and third have a relative angle of 90
while the
middle polarizer has an angle relative to the rst (and therefore an angle 90 relative to the
third). If the intensity after passing through the rst polarizer is I
1
, and the intensity after passing
through the second polarizer is I
2
, then Malus law gives:
I
2
= I
1
cos
2
(13.2)
The intensity after the third polarizer is found similarly, using the relative angle between the second
and third polarizers:
I
3
= I
2
cos
2
(90 ) = I
1
cos
2
cos
2
(90 ) (13.3)
Using the trigonometric identities cos ( )=cos cos +sin sin , and 2 cos sin =sin 2 we
can simplify this result:
I
3
=
1
4
I
1
sin
2
2 (13.4)
This is the curious aspect: if we had only the rst and third polarizers, which are crossed, the
transmitted intensity would be zero. By inserting an intervening polarizer at an intermediate
angle, we recover a non-zero intensity! This clearly demonstrates the vector nature of the electric
(and magnetic) eld. If the intervening polarizer is placed with an angle of 45
, we see that the

transmitted intensity is 1/4 the original intensity.
13.1.2 Photoresistors
A photoresistor, also called a light-dependent resistor or cadmium sulde (CdS) cell, is a resistor
whose resistance decreases with increasing incident light intensity. Photoresistors are made of a
high resistance semiconductor. If the incident light is of sucient energy to excite electrons in the
semiconductor from the valence to conduction band, the resulting free electron and its hole partner
are available to conduct electricity, thereby lowering the resistance. In extrinsic devices, impurities
(or dopants) are added which introduce energy levels nearer the conduction band. This means the
electrons may be excited with even less energy, and lower energy (longer wavelength) photons can
also inuence the resistance. Cadmium sulde, the usual photoresistor material, has a bandgap of
2.42 eV, corresponding to green light of wavelength 512 nm. Thus, CdS photocells have greatest
sensitivity to green light, and their spectral response in fact closely mimics that of the human eye.
A typical photoresistor is shown in Fig. 13.3.
Figure 13.3: A light-dependent resistor. From http://en.wikipedia.org/wiki/Photoresistor.
In the present experiment, you will shine a laser beam through the polarizers under study and
measure the intensity of the transmitted light by measuring the resistance of a photoresistor.
One complication is that the response of a photocell is a non-linear function of incident light inten-
sity. That is, if the incident light intensity doubles, the resistance will almost certainly not change
by a factor of 2. You will need to calibrate your photocell by rst varying the intensity of the laser
in a known fashion using a variable transmission lter, and measuring the resistance versus percent
transmission. Once this characteristic is known, subsequent measurements of resistance (using the
same source of illumination) can be converted into accurate measurements of transmission percent-
age.
Typically, for a photoresistor the resistance R exhibits a power law characteristic as a function of
incident intensity. If the full (unblocked) intensity of the source is I
o
, and the resistance at that
intensity is R
o
, the behavior should follow the form
R
R
o
= a
_
I
I
o
_
b
(13.5)
where a 1 and b 0.5 are constants to be determined for your photoresistor. After measuring
your R(I) characteristic, you will perform a least-squares regression to determine the coecients for
the power law above. This will calibrate your photocell and allow relative intensity to be accurately
determined from a resistance measurement.
13.1.3 Diraction Grating
You may want to review the basics of single- and multiple-slit diraction before reading this section,
which can be found in essentially any introductory electricity and magnetism or optics text.
13.2 Objective 127
A diraction grating is an optical component with a regular pattern, in this case many thousands
of parallel lines. The form of the light diracted by a grating depends on the structure of the
elements and the number of elements present, but all gratings have intensity maxima at angles
n
which are given by the grating equation
d (sin
n
+ sin
i
) = n. (13.6)
where
i
is the angle at which the light is incident, d is the separation of grating elements and
n is an integer which can be positive or negative. The light diracted by a grating is found by
summing the light diracted from each of the elements, and is essentially a convolution of diraction
and interference patterns. In the present experiment, you will measure the diraction angle and
use the known grating spacing to determine the wavelengths of the lasers used in the preceding
experiments.
13.2 Objective
Verify the dependence of transmitted light intensity on the angle between two or three polarizers
placed in series. This demonstrates the basic functionality of a liquid crystal display (LCD), where
the central polarizer is a twisted nematic liquid crystal whose polarization can be controlled by
the application of a small voltage. Polarization also plays a crucial role in some newer 3D movie
technologies.
Verify that the wavelength of light may be measured by diraction through a grating of closely-
spaced lines. You will use this technique in later spectroscopy labs.
Hypothesis: The intensity of light passing through two polarizers is proportional to the square
of the cosine of the relative angle between the polarizers axes. A simple extension of this model
gives the intensity after passing through three polarizers. The position of diracted monochromatic
light beams after passing through a grating of a known spacing of vertical lines may be used to
determine the wavelength of the light.
1. For 3 polarizers, what is the angle between the middle polarizer and the rst polarizer to get
the maximum transmission through all 3 polarizers?
2. For an incident wavelength of =632 nm, what do you expect the angle to be for the rst
diracted beam with a diraction grating of 600 lines/mm?
128 13.4 Supplies
Taylor[1], Ch. 2, 3
Review of basic wave optics (e.g., Serway, Halliday & Resnick, etc.)
http://en.wikipedia.org/wiki/Polarizer
http://en.wikipedia.org/wiki/Polarization_(waves)
http://en.wikipedia.org/wiki/Diffraction
13.4 Supplies
1. Red diode (650 nm) and green (=543 nm) laser
2. 3 Polarizers (with angular scale)
3. 1 Variable transmission lter
4. 48 mm or 22 mm convex lens
5. 1 Diraction grating
6. 1 Diraction optics wheel
7. 3 Magnetic optical component holders
8. Magnetic optical rail
9. 1 Photoresistor, mounted to optical component holder
10. Digital voltmeter
11. Safety goggles for red & green lasers
12. 2 banana cables
13. lab notebook
The rst task is to set up the optical rail system and collect the relevant components listed above.
13.5.1 Setup
1. Once you have the required items, place the laser on one end of the magnetic optical rail.
You may choose either the red or green laser for the polarization experiment. The red diode
laser has built-in positioning, which may be more convenient.
2. At all times, BE CAREFUL NOT TO DIRECTLY VIEW THE BEAM OR ITS DIRECT
REFLECTION. Use the green safety googles with the red laser, and vice-versa.
3. Plug the laser in to the nearest outlet, taking care that the cord is secure and will not be
disturbed during your measurements aside from the obvious safety issue of tripping over
loose cords, vibrations in the cord will be coupled into your optical system and make your
readings more variable. Vibration isolation is important! Do not turn on the laser yet.
4. Place the photodiode in its component holder on the opposite end of the rail.
5. Connect the photoresistor to the digital voltmeter using the banana cables (polarity is not
important), again taking care to secure the cables against vibrations.
6. Turn on the laser. Do not, under any circumstances, stare into the beam or point it into your
eye directly. Though the lasers we are using are not dangerous per se, extended exposure can
still cause vision impairment. Align the laser spot such that it falls directly on the center of
the photocell.
7. Turn on the digital voltmeter, and set it for a resistance (ohms) measurement. Set the
voltmeter range at 20 k, and do not change the range throughout the experiment.
ii
The
resistance of your photoresistor under full illumination of the laser should be approximately
0.51 k
8. You may wish to use a short focal length lens to focus and steer the beam onto the pho-
toresistor; a 48 mm convex lens works well. If you choose to use a lens, leave it in place for
all measurements to maintain your calibration.
13.5.2 Calibrating the Photoresistor
1. Place the variable transmission lter in the path of the laser beam using an optical component
holder, such that the beam passes through the 0% transmission region. Place the lter as
close to the photoresistor as possible, as the scattering by the lter will spread the beam.
2. Record the resulting resistance of the photoresistor.
iii
3. Move the variable transmission lter such that the 20% transmission window intercepts the
beam, and record the resistance. Repeat for 40, 60, 80, and 100% transmission and complete
the table of transmission and the corresponding photoresistance below. Include the estimated
uncertainty in the resistance.
4. Make a plot of your resistance (y) versus transmission (x) data in, e.g., Excel and perform a
power-law regression (termed a trendline in Excel). Alternatively, you may make a table of
log R versus log T and perform a linear regression, which can be performed on most handheld
calculators. Do not include the 0% transmission point in your trendline analysis . . . think
about that for a minute.
5. Verify that your t function predicts approximately the correct resistance for a given trans-
mission percentage, and note the t parameters and R
2
value.
If you wish to obtain a more accurate calibration, there are two simple suggestions: rst, use a
second transmission lter in series to have more calibration points; second, shield as much ambient
ii
At a xed range, the voltmeter performs a resistance measurement by sourcing a known current and measuring
the resulting voltage. Lower resistance ranges use higher currents. Since your semiconductor photoresistor is a
non-ohmic device, changing the current will change the resistance value. Thus, if you change the range during your
measurements, you will not be able to compare all of your data together.
iii
You will also be measuring the ambient light as well, which we are assuming is constant. For this to be true,
you should be careful not to read the resistance while, e.g., your shadow is covering the photoresistor. Make each
measurement in the same way as much as possible.
Table 13.1: Photoresistor calibration
Transmission R (k)
0.20
0.40
0.60
0.80
1.00
light as possible, both during the calibration and following experiments. For the latter suggestion,
shielding the region surrounding the detector is sucient, e.g., using a small box or tube.
13.5.3 Two Polarizers
Now you are ready to verify Malus law.
1. Place a single polarizer in the beam path. Leave the transmission lter in place near the
photodiode such that the beam passes through the now-calibrated 100% transmission region
for all measurements.
iv
2. Verify that the beam is still falling on the photoresistor.
3. Record the photodiode resistance. This corresponds to the intensity after passing through a
single polarizer, i.e., I
o
.
4. Place the second polarizer in the beam path, such that each polarizer about 1/3 of the way
along the space between the laser and photoresistor. Again, leave the transmission lter in
place at 100%.
5. Set the both polarizers to 0
.
6. Align the elements by checking that the beam reected from each element falls near the center
of the previous element.
7. Verify that the beam is still falling on the photoresistor.
8. Record the resistance, which corresponds to the light intensity.
9. Vary the angle of only one polarizer in 10
increments up to 180
and record the resistance

values.
10. Record a few extreme points at, e.g., 350 and 190
to ensure that the minima and maxima

are well-located.
11. Using your calibration function for the photoresistor, calculate the relative transmission cor-
responding to each resistance and complete the table below.
iv
Even the so-called 100% transmission region results in some loss in intensity due to scattering by the plastic lm.
Leaving the lter in at 100% insures that your calibration above remains valid for absolute intensities.
Table 13.2: Intensity versus relative angle between polarizers
(
) R (k) Transmission
single pol.
350
0
10
. . .
170
180
190
13.5.4 Three Polarizers
1. Set the rst polarizer to 0
and the second to 90
. Place a third polarizer between the two

polarizers you already have. Again, leave the transmission lter in place for 100% calibrated
transmission.
2. Repeat the procedure above, varying the angle of only the central polarizer.
You can now remove all optical components from the rail except for the laser.
1. Place the wooden block with graph paper at the far end of the rail. Center the laser beam
on the center of the graph paper.
2. Insert the diraction grating (with a magnetic mount) between the laser and graph paper.
Adjust its horizontal position until you see a central laser spot and two diracted spots on
either side of it.
3. Align the grating and laser such that the incident beam is normal on the grating (i.e.,
i
=0).
An easy way to do this is to check that the beam reected from the grating hits the center of
the laser.
4. Carefully measure the positions of the central spot and the two diracted beams.
5. Carefully measure the distance between the grating and the graph paper.
6. Move the grating close enough to observe two more sets of diracted spots (for a total of 5
spots, including the central beam). Measure the spot positions.
7. Switch to the other color laser and repeat the measurements above.
8. Qualitatively explore the patterns resulting when you project the laser beam through the
various features on the diraction optics wheel.
If there is time: Using the photoresistor, measure the intensity of the central and rst two pairs of
diracted spots.
13.6 Data analysis
13.6.1 Two Polarizers
1. Plot the relative transmission versus angle, including error bars corresponding to the propa-
gated error due to resistance uctuations and your measurement of the polarizer angles.
2. On the same plot, include a calculated response from Malus law, varying the parameters
to determine the initial intensity and any angular oset until you achieve a satisfactory t.
Minimizing
2
is a good way to achieve this!
13.6.2 Three Polarizers
Repeat the analysis above, including a calculated response from Eq. 13.4.
13.6.3 Diraction
From the positions of the spots and the distance to the grating, calculate the sine of the angle dened
by the beam incident on the grating, the central beam on the graph paper, and the diracted spots.
If the optical elements are aligned for normal incidence of the beam on the grating, the n
th
spots
away from the central beam should have a position governed by
n = d sin (13.7)
where n is the order of the spot (n=1 for the spots immediately following the central beam, is
the incident wavelength, and d is the grating spacing (i.e., the inverse of the lines per meter of
the grating). From your distance measurements, determine for each laser, including the sources
and magnitude of the uncertainties.
reports).
1. How does the three-polarizer experiment relate to a liquid crystal display (LCD)? How can
you prove that your cell phone display uses liquid crystal technology? Hint: try viewing it
through a polarizer!
2. Why are polarized sunglasses useful?
3. Explain how polarization is used in some recent 3D movie technologies.
4. Does the order of the polarizers in the three polarizer experiment matter? Why?
5. How are dierent sorts of common polarizers constructed? What is the mechanism for pro-
ducing polarized light?
6. How accurate is this technique of wavelength measurement? How could you improve it?
7. For what other experiments in this course will the diraction measurement have relevance?
8. You can observe diraction in a household setting with a simple laser pointer. A good
grating is a CD or DVD disk. Why?
9. Discuss qualitatively the patterns resulting from the features on the optics wheel. Why do
they appear the way they do?
14
Fine Structure in Atomic Spectra
14.1 Introduction
In this experiment, we will employ a plane diraction grating, collimating optics, and a telescope
to form a spectrometer, which we will use to measure a visible atomic emission lines of hydrogen
(H), deuterium (
2
1
H) sodium (Na) and mercury (Hg). We will use the Hg lines to calibrate the
spectrometer, and subsequently (1) estimate the Rydberg constant for hydrogen, (2) measure the
dierence in the hydrogen and deuterium spectrum, and (3) determine the energy level diagram of
the low lying levels of Na and the spin-orbit splitting of the 3p conguration.
In this experiment, we will use a plane diraction grating, the large N limit of diraction by
multiple parallel slits. Diraction gratings can be made by ruling a large number of equidistant
lines on a glass substrate. Most gratings today, however, are replicas (molds of ruled gratings).
Parallel light rays incident on a diraction grating produces a series of sharp spots at varying angles
with respect to the incident beam. For parallel light incident on a diraction grating with angle
i
, the positions of the maxima are given by
p = d (sin
p
+ sin
i
) (14.1)
where d is the spacing between lines of the grating,
p
the diraction angle,
i
the angle of incidence
with respect to the grating normal, and p designates the order of the spectrum (see Fig. 14.1). It is
clear from Eq. 14.1 that for xed p, the angle
p
will be a function of the wavelength of the light .
Thus, if we illuminate the grating with parallel rays of light composed of several wavelengths, each
wavelength will emerge with a dierent angle. Note, however, dispersion only occurs for [p[ 1. At
zero order,
p
=0, we will see all the wavelengths.
The diraction grating is a particularly simple instrument to use for studying the components of
light produced by an excited atom. When the electrons in an atom are excited, they return to lower
states by emitting light of specic wavelengths known as spectral lines. The spectrum of an atom
is one of its basic identifying signatures. The existence of many elements in astronomical studies is
often established by measurements of their spectra. Atomic spectroscopy has also played a funda-
mental role in the development of quantum mechanics, and the great theoretical and experimental
importance of atomic spectroscopy is well established.
135
d
collimated light
incident on grating
grating
p
p=0
p=-1
p=-2
p=1
p=2
i
Figure 14.1: Schematic illustration of diraction of light by a diraction grating. Various orders of the spectrum are shown.
14.1.2 Hydrogen Atom
original levels electromagnetic radiation is emitted. The resulting light intensity versus wavelength
for an excited gas should thus contain discrete peaks, corresponding to discrete atomic transitions
unique to individual atoms. Some of this radiation may be in a wavelength region that is visible
to the human eye. Similarly, when light passes through a gas, photons whose energies correspond
to these transitions will be absorbed, leading to a loss of intensity at the corresponding wavelengths.
In this experiment, a vapor (such as mercury or hydrogen) is placed in an electric discharge tube
and a high voltage is placed across the tube. The excited emission may look almost white or have
a characteristic color depending on the vapor inside, but it is in reality composed of a number of
dierent colors or wavelengths of visible light. For the purposes of discussion, we will focus now on
the hydrogen atom.
A series of lines in the emission spectrum of excited hydrogen gas is the set of spectral lines observed
when various excited states of the hydrogen atom decay into a common nal state, illustrated
schematically in Fig. 14.2. In the case of the Balmer series, which contains several wavelengths in
the visible range, this nal state is the n = 2 state of hydrogen. The quantity n is the so-called
principal quantum number of the hydrogen atom. The energy levels of the hydrogen atom are given
by the formula
E
n
=
E
i
n
2
n N
1
= 1, 2, 3, . . . E
i
= ionization energy = 13.6 eV (14.2)
Essential to the derivation of emission and absorption spectra is the postulate that a single light
quantum, more commonly called a photon, of energy E
n
=hf =hc/ is emitted when the atom
Figure 14.2: In the simplied Bohr model of the hydrogen atom, the emission lines correspond to an electron jump from a
higher energy level n to a lower n
with the emission of a photon corresponding to the energy dierence between the two levels.
For the visible Balmer series, n
=2 and n runs from 3 upward. From http://hyperphysics.phy-astr.gsu.edu/.

makes a downward transition from an excited state to a lower state (here h is Plancks constant, c
is the speed of light, f the frequency of emitted light, and the wavelength of emitted light). The
energy of the light emitted from a transition between two energy levels E
n
and E
n
is thus equal
to the photon energy,
hf =
hc
= E
n
= E
n
E
n
= E
i
_
1
n
2

1
n
2
_
(14.3)
Hence, for the case of the Balmer series where the nal state n
=2 is the same for every line, we

obtain
hf =
hc
= E
n
= E
n
E
n
= E
i
_
1
2
2

1
n
2
_
(14.4)
Other series include the Lyman series (nal state n = 1, emission in the UV) and the Paschen
series (nal state n=3, emission in the infrared). Some of the lines from the Paschen series can be
observed with the USB-controlled spectrometer you will use. Often the spectral lines are labeled
, , , , etc., starting with the longest wavelength line.
For historical reasons, the formula for hydrogen emission lines is often given in terms of the principle
quantum numbers, the emission wavelength, and the Rydberg constant R
:
1
= R
_
1
n
2

1
n
2
_
(14.5)
R
=
m
e
e
4
4c
3
(14.6)
This presumes the nucleus to be innitely massive (and thus at rest), clearly an approximation.
For an atom whose nucleus has mass M, the resulting reduced mass of the atomic system alters
the Rydberg constant:
R
M
= R
_
1 +
m
e
M
_
1
(14.7)
This correction due to the reduced mass of the atomic system is particularly important when com-
paring the hydrogen and deuterium spectra. Your measurements of emission lines of hydrogen and
deuterium will ultimately give you an experimental value for the Rydberg constant, which you can
compare to accepted values. A derivation of this relationship is provided in the appendix.
14.1.3 Sodium Atom
The structure of Na is more complicated than that of H because of the presence of 10 additional
electrons. Although these electrons are tightly bound in the 1s, 2s and 2p levels, they produce a
signicant aect on the energy of the outer electron in the 3p level.
The well known bright doublet which is responsible for the bright yellow light from a sodium lamp
may be used to demonstrate several of the inuences which cause splitting of the emission lines of
atomic spectra. The transition which gives rise to the doublet is from the 3p to the 3s level, levels
which would be the same in the hydrogen atom. The fact that the 3s (orbital quantum number = 0)
is lower than the 3p (l =1) is a good example of the dependence of atomic energy levels on angular
momentum. The 3s electron penetrates the 1s shell more and is less eectively shielded than the
3p electron, so the 3s level is lower (more tightly bound). The fact that there is a doublet shows
the smaller dependence of the atomic energy levels on the total angular momentum . The 3p level
is split into states with total angular momentum j =3/2 and j =1/2 by the magnetic energy of the
electron spin in the presence of the internal magnetic eld caused by the orbital motion. This eect
is called the spin-orbit eect. In the presence of an additional externally applied magnetic eld,
these levels are further split by the magnetic interaction, showing dependence of the energies on
the z-component of the total angular momentum. This splitting gives the Zeeman eect for sodium.
The magnitude of the spin-orbit interaction has the form
z
B =
B
S
z
L
z
. That is, the spin-orbit
interaction can be regarded as an internal magnetic eld which raises or lowers the energy of
electrons according to their spin. In the case of the sodium doublet, the dierence in energy for
the 3p
3/2
and 3p
1/2
comes from a change of 1 unit in the spin orientation with the orbital part
presumed to be the same. The change in energy is of the form
E =
B
gB (14.8)
where
B
=57.88 eV/T is the Bohr magneton and g 2 is the electron spin g-factor. This gives
an estimate of the internal magnetic eld needed to produce the observed splitting.
14.1.4 Mercury Atom
The structure and spectra of the Hg is very complicated, due to the large number of electrons
(Z =80). Mercury is added to uorescent lights because the large number of dierent wavelength
lines in the spectra yields an overall color close to white. We will use the most intense mercury
lines to calibrate the spectrometer, and investigate its spectrum more qualitatively.
Figure 14.3: A digital picture of the Hg light source taken with a diraction grating placed over the lens.
14.1.5 The Grating Spectrometer
The apparatus used in this experiment is schematically shown in Fig. 14.4. It consists of a plane
transmission diraction grating (600 lines/mm and 1200 lines/mm gratings are available) mounted
on the rotating table of a spectrometer. The spectrometer also consists of a telescope and a
collimator, both of which can be rotated independently about the axis of the table. A description
of each of the numbered parts will now be given.
1. Table. Used for mounting the grating.
2. Eyepiece focus. At the end of the telescope, this eyepiece can be gently pulled or pushed to
get a sharp focus on the cross- hairs without moving their position. This allows the apparatus
to allow for dierent individual viewing preferences without up- setting the focus of the image
being viewed.
3. Telescope focus. The chrome ange can be rotated to focus the telescope on the image.
4. Leveling Screws. These are used to align the collimator and telescope vertically . One screw
is a clamp and the other is used for leveling.
(1)
(4)
(4)
(3)
(2)
(7)
(8)
(8)
(5)
(6)
(10)
(9)
Figure 14.4: The experimental apparatus used to study atomic spectra. The monkey is optional.
5. Collimator stop. Adjusts the length of the collimator. The collimator should be set to the
correct length when the stop is pushed in against the outer shell.
6. Collimator slit. Allows adjustment in width and orientation of the slit.
7. Telescope arm clamping and tangent screws. The two screws on the side of the spectrometer
just below the telescope each serve a distinct purpose. The screw projecting radially out-
ward is the clamping screw which should always be loosened before attempting to move the
telescope. After the telescope is nearly in the desired position, the clamping screw should be
gently tightened. The perpendicular tangent screw then allows for very ne adjustment of
the telescope position.
8. Object table clamping and tangent screws. These adjust the rotating table in the same way
as (7) adjusted the position of the telescope arm. They should be used in the same way.
9. Angular position readout, with magnier. The least division of the main scale is 1/3 degree
and there is a vernier to carry the measurement to 1 minute of arc.
10. Diraction grating.
Below is a simplied schematic of the key optical elements of the spectrometer, showing the prin-
ciple of operation.
14.1.6 Aligning the Spectrometer
First, focus the telescope for parallel light (i.e., light from innity) by focusing on an object several
tens of meters away. Focusing through a window on a building across the street will do. Then,
minimize the parallax between the focused image and the cross-hairs. That is,locate the image in
the plane of the cross-hairs.
Second, put a light source close to the slit of the collimator. Using the telescope, view the slit.
Make the slit as narrow as possible. While looking through the telescope, adjust the focus of the
collimator so that the image of the slit falls on the plane of the cross-hairs. This is done by remov-
collimator
slit
collimator
parallel
beam
grating
r
e
d

l
i
g
h
t
green light
diffraction angle
te
le
s
c
o
p
e
e
y
e
p
ie
c
e
light
source
Figure 14.5: Principle of operation of the spectrometer.
ing any parallax between the image of the slit and the cross- hairs. Now the collimator is set to
produce a collimated beam, i.e., a parallel beam of light.
Third, mount the grating on the table of the spectrometer. Make sure that the photographic lm
from which the grating is made is on the telescope side. (The excess glass should protrude towards
the telescope side if you use the 1200 lines/mm grating.) Note the position of the telescope when
it is in line with the collimator, i.e., when the slit image is aligned with the cross-hairs. Shine the
light from your desk lamp into the open side port of the telescope (near the eyepiece) to illuminate
the cross-hairs. You might nd it helpful to place a mirror just behind and ush with the grating
to see the reected image of the cross-hairs. Once you have located it, you can remove the mirror
and continue with the alignment procedure. Rotate the grating so that the reected image of
the cross-hairs falls back on the cross-hairs. In this position, the grating is perpendicular to the
optical axis of the telescope. The system is set up to approximately satisfy Eq. 14.1; the incident
beam consists of parallel light falling nearly normally on the grating, i.e.,
i
=0. Lock the grating
into place and record the position of the angular scale on the rotating table. Do not adjust the
grating anymore. You will be able to see images of successive orders of the spectrum by rotating
the telescope away from the zero order position.
14.1.7 Making Measurements of Various Spectra
As mentioned earlier, the sources you will be observing are mercury, hydrogen, and sodium. You
will be able to see rst order spectra for all of these light sources. With the spectrometer enclosed
in the dark box, you will also be able to detect some of the lines in second order.
To make measurements, rst record the value of the angular scale on the rotating table when the
cross-hairs are aligned with the zero order spectrum (this corresponds to
o
). Rotate the telescope
142 14.2 Objective
Figure 14.6: Green, yellow, and red Hg emission lines as viewed through the spectrometer. Note that all three lines are
actually doublets.
to the right so that the cross-hairs align with a p
th
order spectral line. Record the angular scale
reading. Then, rotate the telescope so that the cross-hairs align with the corresponding spectral
line on the left side of the zero order spectrum; record the angular scale reading. You should
measure the angle for each spectral line at least twice on each side, giving you at least
four measurements of the angle,
p
for the p
th
order. Take the average as your result and the
standard deviation as your uncertainty. This should be done for each visible spectral line. Note,
you might nd it easier to measure all the lines on one side before measuring them on the other
side. With these measurements, you will be able to calculate , the wavelength of each spectral line.
The scale on the spectrometer is a vernier scale. The outside scale is ruled in 0.5 degree; the inside
scale is ruled in minutes of arc. To read the angle, look at 0 on the inner scale. Say it falls between
180.0 and 180.5 on the outer scale. Now look for the point where the outer and inner scales line
up. Say it is aligned with 24 on the inner scale. Your angle would be 180
24
. If the 0 of the inner

scale were between 180.5 and 181.0 then your angle would have been 180
54
.
Make sure you understand how to read a vernier scale. A nice applet illustrating how to read
the vernier scale can be found at: http://www.phys.hawaii.edu/%7Eteb/java/ntnujava/ruler/
vernier.html. When making your measurements, always rotate the table in the same direction,
cw or ccw, to minimize errors due to hysteresis.
14.2 Objective
Measurement of the ne structure in atomic spectra in H and Na, particularly eects due to the
reduced mass of the atomic system and the spin-orbit interaction.
14.3 Supplies & Equipment 143
1. H, D vapor discharge tubes
2. Vapor lamp housing & power supply
3. Na and Hg high-pressure lamps
4. Diraction grating (600 or 1000 lines/mm) and spectroscope
5. lab notebook
14.4.1 Mercury Spectrum
You will use the Hg spectrum as your wavelength standards to calibrate your spectrometer. The
wavelengths can be on the National Institute of Standards and Technology atomic database website:
(http://physics.nist.gov/PhysRefData/ASD/index.html) or in the attached handout. You
should measure the angles of the ve lines (violet, blue, blue-green, green and yellow). From the
known wavelengths, you can use Eq. 14.1 to rene your value of d (1/600 mm). Do this by tting
to d sin and extracting a value for d from the t parameters.
In addition to measuring the ve most intense lines, note the presence of any additional lines,
particularly doublets and triplets. Can you index all lines to known atomic transitions? What
eects are responsible for the various doublets and triplets?
14.4.2 Hydrogen Spectrum
With the hydrogen lamp, measure the wavelengths of the violet, blue and red spectral lines. The
wavelengths you measure are the air wavelengths. To determine the Rydberg constant, you will
need vacuum wavelengths. Given the air wavelength, the vacuum wavelength can be calculated
with
vac
= n
air
air
(14.9)
where n
air
is the index of refraction of air; its value can be found in the Handbook of Physics and
Chemistry or on the web. You can determine R
H
by tting 1/
vac
to
R
H
_
1
2
2

1
n
2
_
(14.10)
Convert R
H
to R
and compare your estimate with the tabulated value.

14.4.3 Deuterium Spectrum
Once you have nished the Hydrogen spectrum, align the crosshairs on the red Hydrogen line.
Without moving anything, replace the Hydrogen lamp with the Deuterium lamp. You should now
observe a similar red line, but in a slightly dierent position (i.e., no longer aligned with the
crosshair). Carefully measure the deuterium spectrum, with a special emphasis on determining the
splitting of the red line between hydrogen and deuterium. Compare this splitting to that predicted
by the reduced mass correction above.
14.4.4 Sodium Spectrum
With the sodium lamp, you should measure the wavelengths of the green, yellow and red spectral
lines in rst order, and compare to the database values. Also note the position of any other lines
(there should be several more, if the ambient light is suciently reduced). Can you explain which
transitions are responsible for all lines?
You should also measure the yellow Na lines in 2nd order if possible. Calculate the splitting of the
yellow doublet in electron volts, and compare to the database values. What is the eective strength
(in Tesla) of the spin-orbit interaction?
Appendix 1: Reference Spectral Data
Table 14.1: Hydrogen emission lines, UV-Vis-NIR[14]
383.5384 5 Balmer 9 2
388.9049 6 Balmer 8 2
397.0072 8 Balmer 7 2
410.174 15 Balmer 6 2
434.047 30 Balmer 5 2
486.133 80 Balmer 4 2
656.272 120 Balmer 3 2
656.2852 180 Balmer 3 2
954.597 5 Paschen 8 3
1004.94 7 Paschen 7 3
Spin-orbit split lines, 2p

1/2
1s and 2p
1/2
1s
Appendix 2: Angular Momentum Correction to Hydrogen Levels
The energies of the stationary states of hydrogen slightly depend on the orbital angular momentum
quantum number l. An improved formula for the energy of the state of quantum numbers n and l
for nonzero l is
E
n,l
=
m
e
e
4
2 (4
o
)
2
2
n
2
_
1 +
e
4
(4
o
)
2
2
c
2
n
_
1
1 + l +
1
2

1
2
3
4n
__
(14.11)
Table 14.2: Selected Sodium emission lines, UV-Vis-NIR[14, 15]
466.5 19 6
2
D
5/2,3/2
3
2
P
1/2
466.9 19 6
2
D
5/2,3/2
3
2
P
3/2
474.8 4 7
2
S
1/2
3
2
P
1/2
475.2 4 7
2
S
1/2
3
2
P
3/2
497.9 104 5
2
D
5/2,3/2
3
2
P
1/2
498.3 104 5
2
D
5/2,3/2
3
2
P
3/2
514.8 23 6
2
S
1/2
3
2
P
1/2
515.4 23 6
2
S
1/2
3
2
P
3/2
568.3 130 4
2
D
5/2,3/2
3
2
P
1/2
568.8 130 4
2
D
5/2,3/2
3
2
P
3/2
589.0 139 3
2
P
3/2
3
2
S
1/2
589.6 139 3
2
P
1/2
3
2
S
1/2
615.4 67 5
2
S
1/2
3
2
P
1/2
616.1 5
2
S
1/2
3
2
P
3/2
818.3 90 3
2
D
5/2,3/2
3
2
P
1/2
819.48 129 3
2
D
5/2,3/2
3
2
P
3/2
Table 14.3: Selected Mercury emission lines, UV-Vis-NIR[14, 16]
365.02 2800
365.48 300
366.29 80
404.66 1800
407.78 140
433.92 250
434.75 400
435.83 4000
546.07 1100
567.59 160
576.68 240
578.97 100
579.07 280
580.38 140
614.95 1000
671.64 160
690.75 250
708.19 250
709.19 250
where the term
1
2
corresponds to the spin parallel and antiparallel, respectively, to the orbital
angular momentum.
Appendix 3: Center-of-Mass Correction to the Hydrogen Energies
In a simple calculation of the energies of the stationary states of hydrogen, we pretend that the
proton remains at rest. Actually, both the electron and proton orbit about their common center of
mass. We can show that the energies of the stationary states, taking into account this motion of
the proton, are given by
E
n
=
me
4
2 (4
o
)
2
2
n
2
(14.12)
where m is the reduced mass
m =
m
e
m
p
m
e
+ m
p
(14.13)
There is a fairly good chance that youve not encountered the reduced mass in your previous me-
chanics classes, so we will go through it quickly. Essentially, the reduced mass is an eective
inertial mass that can be used solve a two-body problem as if it is a single body problem.
Our system consists of a single proton and a single electron, separated by a distance r described by
a vector r . Classically, we expect both particles to orbit about their center of mass. The proton,
being much heavier, will move very little in comparison to the lighter electron, and sit closer to the
center of mass. This is depicted below.
i
The center of mass for a simple two-body system like this essentially amounts to nding the point
along their common axis at which their torques would balance.
ii
If the particles positions are
described by vectors r
e
and r
p
pointing from the center of mass toward the electron and proton,
respectively, then this requires
m
e
r
e
+ m
p
r
p
= 0 or r
p
=
m
e
m
p
r
e
(14.14)
Of course, we also know that the r =r
e
r
p
. Rearranging the rst equation and substituting,
r = r
e
r
p
= r
e
_
1 +
m
e
m
p
_
= r
p
_
m
e
m
p
+ 1
_
(14.15)
This leads us to expressions for the electron and proton orbital radii in terms of their separation
and masses:
r
e
= r
_
m
p
m
p
+ m
e
_
r
p
= r
_
m
e
m
p
+ m
e
_
(14.16)
Here the negative sign for r
p
just indicates that the direction is opposite r . Two other interesting
things to note: rst, only the relative mass m
e
/m
p
really matters; second, owing to the fact that
i
Some neat animations can be found at http://en.wikipedia.org/wiki/Barycenter.
ii
Put more simply, if the two particles were on a see-saw, where should the center pivot be?
e
p
+
r
r
e
r
p
CM
X
Figure 14.7: An electron and a proton, with a separation r . The electron sits at a distance re from their center of mass, and
the proton a distance rp from the center of mass.
m
p
m
e
, the proton will orbit in a much smaller circle than the electron. So much smaller, in fact,
that we are usually well-justied in taking the proton to be at rest. Now, let us write down the
kinetic energy of the two-particle system. Well have the kinetic energy of each particle, and their
electrical potential energy :
E =
1
2
m
e
v
2
e
+
1
2
m
p
v
2
p
k
e
e
2
r
=
1
2
m
e
_
dr
e
dt
_
2
+
1
2
m
p
_
dr
p
dt
_
2
k
e
e
2
r
(14.17)
Using the expressions above, we can write the individual particles velocities in terms of the rate
at which the separation changes with time:
v
e
=
dr
e
dt
=
_
dr
dt
_
_
m
p
m
p
+ m
e
_
v
p
=
dr
p
dt
=
_
dr
dt
_
_
m
e
m
p
+ m
e
_
(14.18)
But dr /dt is nothing more than the relative velocity of the proton and electron, which well simply
call v:
v =
dr
dt
=
d
dt
(r
e
r
p
) =
dr
e
dt

dr
p
dt
(14.19)
With this denition, we may write the electron and proton velocities as
iii
v
e
= v
_
m
p
m
p
+ m
e
_
v
p
= v
_
m
e
m
p
+ m
e
_
(14.20)
iii
We dont really need the vector notation any more, since were just going to square them anyway.
Substituting these expressions into our energy equation,
E =
1
2
m
e
v
2
e
+
1
2
m
p
v
2
p
k
e
e
2
r
=
1
2
m
e
v
2
m
2
p
(m
p
+ m
e
)
2
+
1
2
m
p
v
2
m
2
e
(m
p
+ m
e
)
2

k
e
e
2
r
(14.21)
E =
m
e
m
p
2 (m
e
+ m
p
)
2
_
m
p
v
2
+ m
e
v
2
_
k
e
e
2
r
=
1
2
_
m
e
m
p
m
e
+ m
p
_
v
2
k
e
e
2
r
(14.22)
If we dene a quantity m=m
e
m
p
/ (m
e
+ m
p
), known as the reduced mass, our energy equation
is identical to that of a single particle of charge e in an electric potential V =k
e
e/r:
E =
1
2
mv
2
eV (14.23)
We have successfully reduced a two-body problem to a single-body problem. How about the angular
momentum? Taking the center of mass as our origin, the net angular momentum is mvr for each
particle:
L = L
p
+ L
e
= m
e
_
dr
e
dt
_
r
e
+ m
p
_
dr
p
dt
_
r
p
(14.24)
L = m
e
_
vm
p
m
p
+ m
e
__
rm
p
m
p
+ m
e
_
+ m
p
_
m
e
v
m
e
+ m
p
__
rm
e
m
p
+ m
e
_
(14.25)
L =
m
e
m
p
vr
(m
e
+ m
p
)
2
(m
e
+ m
p
) =
_
m
e
m
p
m
e
+ m
p
_
vr = mvr (14.26)
Again, we have reduced a two-body problem to an equivalent single-body problem: the total angular
momentum is simply the reduced mass times the relative velocity times the separation distance.
We can now apply our quantization condition for the total angular momentum:
mvr = n or v =
n
mr
(n N
1
= 1, 2, . . .) (14.27)
Substituting this condition into our energy equation:
E =
1
2
m
_
n
2
2
m
2
r
2
_
k
e
e
2
r
=
n
2
2
2mr
2

k
e
e
2
r
(14.28)
The denition of a stationary state is a state in which the energy has no time dependence. In
this case, there is no explicit time dependence, but an equivalent condition is to require that the
variation with respect to radius vanishes:
dE
dt
=
dE
dr
dr
dt
= 0 =
_
n
2
2
mr
3
+
ke
2
r
2
_
dr
dt
(14.29)
=
n
2
2
mr
= ke
2
or r =
n
2
2
kme
2
(14.30)
Using this extremal radius in our energy equation, we nd
E =
n
2
2
2m
_
k
2
m
2
e
4
n
4
4
_
k
2
me
4
n
2
2
(14.31)
E =
k
2
me
4
2n
2
2
=
me
4
2 (4
o
)
2
n
2
2
(14.32)
For the last line, we noted that k
e
=1/4
o
. This implies that the hydrogen and deuterium
iv
nuclei
will have slightly dierent emission and absorption lines, owing to the fact that the deuterium
nucleus is about twice as heavy as the hydrogen nucleus. The deuterium transitions should be at
slightly higher energies (lower wavelengths). It is a tiny eect, about a 0.2 nm shift of the 656.3 nm
H line, but quite measurable and in good agreement with this simple model.
v
Similar eects are
expected and observed for other hydrogen-like systems, e.g., He
+
and Li
++
.
iv
Deuterium is hydrogen with a neutron in the nucleus in addition to the proton.
v
See, for example, http://hyperphysics.phy-astr.gsu.edu/Hbase/quantum/hydfin.html.
15
Charge Quantization
For this experiment, you will follow the Pasco Scientic procedure, available at ftp://ftp.pasco.
com/Support/Documents/English/AP/AP-8210/012-06123D.pdf (a hard copy is available in
the laboratory). Below is a short introduction, and a somewhat lengthy discussion of forces in
uids relevant to the data analysis for this lab.
15.1 Objective
Demonstrate that electric charge is quantized and determine the base unit of electric charge.
Taylor[1], Ch. 3, 4
1. Lab notebook
2. Millikan oil drop apparatus and accessories (Pasco)
3. VideoFlex microscope & monitor or PC
4. High-voltage power supply (Pasco)
5. Multimeter
15.4 Introduction
The electric charge carried by a particle may be calculated by measuring the force experienced
by the particle in an electric eld of known strength. Although it is relatively easy to produce a
known electric eld, the force exerted by such a eld on a particle carrying only one or several excess
electrons is very small. For example, a eld of 10
5
V/m would exert a force of only 1.6 10
14
N
on a particle bearing one excess electron. This is a force comparable to the gravitational force on
a particle with a mass of 10
15
kg!
i
i
Much of this procedure is reproduced from Pasco document 012-6123D, Instruction manual and experiment
guide for the PASCO scientic Model AP-8210. This partial reproduction is only for use in PH255 at the University
of Alabama, as dictated by the copyright notice in 012-06123D.
151
The success of the Millikan Oil Drop experiment depends on the ability to measure forces this
small. The behavior of small charged droplets of oil, having masses of only 10
15
kg or less, is
observed in a gravitational and an electric eld. Measuring the velocity of the drop falling in air
enables, with the use of Stokess law, the calculation of the mass of the drop. The observation of
the velocity of the drop rising in an electric eld then permits a calculation of the force on, and
hence, the charge carried by the oil drop.
Although this experiment will allow one to measure the total charge on a drop, it is only through
the careful analysis of the data obtained and a certain degree of experimental skill that the charge
of a single electron can be determined. By selecting droplets which rise and fall slowly, one can be
certain that the drop has a small number of excess electrons. A number of such drops should be
observed and their respective charges calculated. If the charges on these drops are integral multiples
of a certain smallest charge, then this is a good indication of the atomic nature of electricity. How-
ever, since a dierent droplet has been used for measuring each charge, there remains the question
as to the eect of the drop itself on the charge. This uncertainty can be eliminated by changing
the charge on a single drop while the drop is under observation. An ionization source placed near
the drop will accomplish this. In fact, it is possible to change the charge on the drop several times.
If the results of measurements on the same drop then yield charges which are integral multiples of
some smallest charge, then this is proof of the atomic nature of electricity.
The measurement of the charge of the electron also permits the calculation of Avogadros number.
The amount of current required to electro-deposit one gram equivalent of an element on an electrode
(the faraday) is equal to the charge of the electron multiplied by the number of molecules in a mole.
Through electrolysis experiments, the faraday has been found to be 9.62510
7
C per kg of equivalent
weight. Dividing the faraday by the charge of the electron yields Avogadros number N
A
:
9.625 10
7
C/kg equiv. weight
1.602 10
19
C
= 6.025 10
23
molecules/kg equiv. weight = N
A
(15.1)
15.4.1 Equation for calculating the charge on a drop
An analysis of the forces acting on an oil droplet will yield the equation for the determination of
the charge carried by the droplet.
First, let us consider the case of a spherical droplet of radius r and mass m falling without an
electric eld present. In this case, there are only two forces acting on the drop: the gravitational
force F
g
pulling it downward, and the frictional (drag) force F
d
pushing upward. The drag force
for very small spherical objects
ii
in a continuous viscous uid can be accurately modeled by Stokes
ii
Strictly, for very small Reynolds numbers.
law
iii
, which gives the drag force as
iv
F
d
= 6rv (15.2)
where F
d
is the frictional (drag) force, is the uids dynamic viscosity, r the radius of the particle,
and v the particles velocity. If the particle reaches terminal velocity, this means that the gravita-
tional force is precisely balanced by the drag force. For a spherical drop of mass m, comprised of a
substance of density , the gravitational force is mg =
4
3
r
3
g, where g =9.81 m/s
2
is the gravita-
tional acceleration of the drop. Figure 15.1a shows the forces acting on the drop when it is falling
in air and has reached its terminal velocity. (Terminal velocity is reached in a few milliseconds for
the droplets used in this experiment.) At the terminal velocity v
f
, the forces must balance:
F
g
= F
d
(15.3)
4
3
r
3
g = 6rv
f
(15.4)
v
f
=
2gr
2
9
(15.5)
Of course, we have in principle neglected one point: the buoyant force due to the diering density
of the drop and the air. Since the density of the air is only about
1
1000
that of oil, this force may
be neglected.
v
F
d
mg
qE
F
d
mg
+ + + + + + + + +
- - - - - - - - -
-
Figure 15.1: (a) This schematic shows the forces acting on the drop when it is falling in air and has reached its terminal
velocity. (b) This schematic shows the forces acting on the drop when it is rising under the inuence of an electric eld.
If the drop in question also has a charge q, then it will experience an additional electric force in an
electric eld. Figure 15.1b shows the forces acting on the drop when it is rising under the inuence
of an electric eld. Here E is magnitude of the electric eld, q the charge on the drop, and v
r
the
iii
Stokes law is derived by solving the Stokes ow limit for small Reynolds numbers (Re1) of the Navier-Stokes
equations.
iv
See Appendix 15.4.1 for a derivation.
v
The correction would involve replacing in the equation above with a, where a is the density of air.
velocity with which the particle rises. Adding the forces, we nd
qE =
4
3
r
3
g + 6rv
r
(15.6)
In principle the radius of the particle can be measured, but it is far easier to measure the terminal
velocity. Based on Eq. 15.5, knowledge of the terminal velocity uniquely determines the particle
radius for a given uid. Eliminating r in favor of v
f
, and solving for q:
qE =
4
3
r
3
g + 6rv
r
= 6r (v
f
+ v
r
) = 6
9v
f
2g
(v
f
+ v
r
) (15.7)
q =
6
E
9
3
2g
(v
f
+ v
r
)
v
f
(15.8)
The electric eld E is generated by applying a potential dierence V between two parallel conduct-
ing plates, placed a distance d apart. If the spacing d is much smaller than the size of the plates in
any lateral dimension (easily satised in the experiment), in the region between the plates the eld
is E=V/d. This lets us eliminate E in favor of the more easily measured V and a known distance.
q =
6d
V
9
3
2g
(v
f
+ v
r
)
v
f
(15.9)
Only one small detail remains: Stokes law becomes inaccurate when the velocity of falling droplets
is less than about 10
3
m/s. Droplets having this and smaller velocities have radii on the order
of 2 m, comparable to the mean free path of air molecules, a condition which violates one of
the assumptions made in deriving Stokess law. Since the velocities of the droplets used in this
experiment will be in the range of 10
4
10
5
m/s, we must add a correction to Stokes law. As
it turns out, the correction is straightforward: we need only replace the viscosity by an eective
value, which includes a correction factor (the Cunningham factor):

eff
=

1 +
b
Pr
(15.10)
Here P is the atmospheric pressure, r is the radius of the drop, and b is a constant factor. This
leads us to a more accurate form for the electric charge on a drop:
q = 6d
_
9
3
2g
_
1 +
b
Pr
_
3
(v
f
+ v
r
)
v
f
V
(15.11)
We can regroup this in a somewhat more convenient manner:
q =
_
_
6d
9
3
2g
_
_
_
1 +
b
Pr
_
3/2
_
(v
f
+ v
r
)
v
f
V
_
(15.12)
The rst quantity in brackets is a function of the experimental apparatus and the composition
of the drop alone, and need only be determined once for any particular apparatus. The second
term in brackets is determined for each droplet, while the third term in brackets is calculated for
each change of charge that the drop experiences. Finally, an alternate form you might nd more
convenient for your calculations is
q =
4
3
gd
_
_
_
b
2P
_
2
+
9v
f
2g

b
2P
_
_
3 _
v
f
+ v
r
V v
f
_
(15.13)
Below, we summarize the denitions of symbols used, along with proper units and numerical values
where applicable.
Symbol Quantity Units Value
q charge coulombs [C]
d separation of plates in condenser meters [m] 7.5 10
3
m
density of oil [kg/m
3
] 866 kg/m
3
g acceleration of gravity [m/s
2
] 9.81 m/s
2
viscosity of air Poise, or [Pa s] see plot
b constant [Pa m] 8.23 10
3
Pa m
P barometric pressure [Pa] 101.325 kPa at 1 atm
r radius of drop [m]
v
f
velocity of fall [m/s]
v
r
velocity of rise [m/s]
V potential dierence between plates [V]
When you are nished
Power o the microscope and video monitor.
Turn the voltage on the accelerating supply to zero, and turn the power o.
Unplug the halogen light from the Millikan apparatus.
Clean up your work area
Viscosity of dry air as a function of temperature
The viscosity of air can be computed using Sutherlands formula
vi
(T) =
o
_
T
T
o
_
3/2
T
o
+ S
T + S
(15.14)
Here is the viscosity in poise (Ns/m
2
) at the input temperature T,
o
is the reference viscosity in
poise (Ns/m
2
) at a reference temperature T
o
, T is the input temperature in Kelvin, T
o
is a reference
temperature in Kelvin, and S is an eective temperature called Sutherlands constant. For dry air,
o
=1.716 10
5
Ns/m
2
at T
o
=273 K, S=111 K, valid over a temperature range of 0555 K.
vii
vi
From http://www.epa.gov/EPA-AIR/2005/July/Day-13/a11534d.htm and http://en.wikipedia.org/wiki/
Viscosity.
vii
Reference data: at 300 K, =1.84610
5
Ns/m
2
. Table A.4 in F.P. Incropera and D.P. DeWitt, Fundamentals of
Heat and Mass Transfer, 3rd ed, John Wiley, New York, NY, 1990. See also table B.4 in D.R. Poirier and G.H. Geiger,
Transport Phenomena in Materials Processing, TMS, Warrendale, PA, 1994)
Derivation of Stokes law and the Navier-Stokes Equation: A crash-
course in uids
Continuity Equation
First
viii
, we need the continuity equation for a uid. Qualitatively, a general continuity equation
reads something like this:
(rate of mass accumulation) + (rate of mass out) (rate of mass in) = 0 (15.15)
We can be a bit more precise by applying our continuity equation to a specic volume of space V ,
which is dened by a bounding surface S. In this case, the net rate at which mass accumulates
inside V depends on the net rate at which mass passes through S, either coming in or going out:
(rate of mass accumulation in V ) + (net rate of mass crossing S) = 0 (15.16)
This is basically just bookkeeping, or conservation of matter if you like. If the amount of mass in
V is static, then it must be true that the amount of matter entering through S is the same as the
amount of matter leaving through S. If the amount of mass in V is increasing, then there must be
a net ow of matter in through S.
For a uid, or any continuous substance, our continuity equation is just an expression of conserva-
tion of matter. Since we wish to deal with continuous substances like uids, rather than particles,
it is most convenient to put our equations in terms of the density of the substance . Consider a
tiny cube of our substance of dimensions xy z. The mass of this cube is simply xy z.
If we have a net ow of our substance through this cube, lets say in the x direction, how does
the mass of the cube change with time? If the substance is incompressible, and the cube remains
completely full, then the mass doesnt change, of course. However, in the general case, we just need
to keep track of how much mass is in the cube at any moment, and how much mass enters and leaves.
We will presume that our cube is nicely aligned along the x, y, and z axes, and that there is a net
ow of our substance with velocity v , as shown in Fig. 15.2. We will assume that the density of
our substance is constant. If we look rst at the faces of the cube perpendicular to the x axis (i.e.,
the faces whose area normals are parallel to the x axis), the net ow through the cube along the x
axis can be found be comparing the rate at which mass enters one side and leaves the other. The
rate of mass owing through the left side face at x is
m
t
x
=

t
(V )
x
=

t
(xyz)
x
= yz (v
x
)
x
(15.17)
viii
Most of this section is based on Ch. 2 of D.R. Poirier and G.H. Geiger, Transport Phenomena in Materials
Processing, TMS, Warrendale, PA, 1994) and Ch. 40-41 of the Feynman Lectures on Physics, vol. II
x
y
z
x
y
z
(x, y, z)
(x +x, y +y, z +z)
(v
x
)|
x
(v
x
)|
x+x
Figure 15.2: Volume element xed in space with uid owing through it.
This is just the familiar result that the mass ow rate through a pipe is product of the velocity of
the ow, the uid density, and the pipes cross-sectional area. In the same manner, we can nd the
ow rate through the right side face at x + x,
m
t
x+x
= yz (v
x
)
x+x
(15.18)
We can proceed similarly for the other two pairs of faces perpendicular to the y and z axes, and then
add up all the terms for uid entering or leaving the cube to come up with a mass balance. If, when
we add up the rates for all the sides, we have a non-zero result, then we must be either accumulating
mass inside our cube, or it is experiencing a net loss in mass. Either way, the accumulation in mass
inside our cube of constant volume can only reect a change in density,
(mass accumulation) =

t
V = xyz
t
(15.19)
Our mass balance is then simply relating this rate of mass accumulation to the net ow through
the cube:
xyz
t
= yz
_
v
x
x
v
x
x+x
_
+ xz
_
v
y
y
+v
y
y+y
_
+ xy
_
v
z
z
v
z
z+z
_
(15.20)
Next, we can divide by xyz and then take the limit of innitesimal dimensions, and after
recalling the denition of the derivative, we arrive at the continuity equation:
t
=
_

x
v
x
+

y
v
y
+

z
v
z
_
= v (15.21)
Hopefully, at least some of this material is familiar. For the present experiment, conducted in
air at very low velocities, we may assume the air has approximately constant density (i.e., it is
incompressible), and the continuity equation simplies to v =0.
Using the vector form of the continuity equation, we can reformulate it for dierent coordinate
systems relevant to specic problems by expanding the divergence operator appropriately. In
spherical coordinates, we have
t
+
1
r
2
r
_
r
2
v
r
_
+
1
r sin
(v
sin ) +
1
r sin
(v
) = 0 (15.22)
Our problem of interest is the slow ow of a uid past a dense sphere. If the uid ow is along
the z axis, the problem is symmetric about the z axis, and we may neglect the components
of velocity. In other words, the problem is essentially two-dimensional, thanks to the rotational
symmetry about the z axis. In this special case,
t
+
1
r
2
r
_
r
2
v
r
_
+
1
r sin
(v
sin ) = 0 (15.23)
As an aside, the equivalent continuity equation in electromagnetism is conservation of charge, which
you might have seen:
/t +
j = 0 (15.24)
where is charge density and

j current density. The charge density in a region can only change if
there is a net ow of charge, a current, into or out of that region.
Static uids
Next, we will need the equation for the forces and momentum in the uid. Let us consider a com-
pletely static volume of uid, with no net ow in any direction. If we know the pressure at some
point within the uid (say, at its bottom surface) is P
o
, then at any point a height h above that
level, the pressure is just P =P
o
gh where g is the gravitational acceleration, and the uid
density. Put another way, the pressure as a depth h diers from our reference level only by the
weight of the uid in a column of height h.
Of course, we can turn this equation around: if P
o
is just an arbitrary, constant reference pressure,
then this also implies that anywhere in the uid P + gh=P
o
must be constant! Actually, this is
not so surprising either, if we multiply everything by a volume of interest:
P
o
V = PV + V gh = PV + mgh (15.25)
What we have is simply a result of the work-energy theorem: the work done in increasing the
pressure on a given constant volume is P(V V
o
), and this must be accounted for by the change
in gravitational potential energy, mgh. Again, in dealing with continuous matter such as a uid, it
is more convenient to recast all of our equations in terms of density rather than mass and volume.
In this light, gh is just the gravitational potential per unit mass, so what we are really saying is
that pressure plus gravitational potential is a constant for a uid, or that pressure itself is a sort of
volumetric potential. Thus, we dene a gravitational potential per unit mass =mg, giving us
P + = const (15.26)
Now we have an energy balance for our static uid, it is only a bit of mathematics to nd a force
balance. If we consider a one-dimensional uid, we know that force is just the spatial derivative
of the potential energy. The same will hold true of the potential energy per unit volume. If we
consider only uids of constant density, we can take the spatial derivative of both sides of Eq. 15.26:
P
x
+
x
= 0 (15.27)
In three dimensions, we need only replace the spatial derivative with a gradient:
P + = 0 (15.28)
This is nothing more than a Newtons law force balance for our stationary uid, if we recognize
that is the force (per unit volume). In static equilibrium, the force per unit volume is precisely
balanced by a gradient in pressure.
This equation is the complete description of hydrostatics, though it is quite a bit more complicated
than it looks: there is no general solution. If the density of the uid varies spatially ( = 0
somewhere), our continuity equation above tells us that there is no way that a static equilibrium
can be maintained, we must have also have time-varying density.
ix
Only if is constant in space
do we have a simple solution for hydrostatics, viz., P + =const.
Equations of motion without viscosity (Dry Water)
What to do if the uid is not static? We already know the continuity equation in general, but we
still need to consider a more general force balance for our uid. What we have derived above is the
equilibrium condition for a static uid, generalizing just means letting the pressure and potential
gradient terms become unbalanced to yield a net acceleration. In the absence of viscous forces, this
would simply be
(acceleration) = P (15.29)
The left side is the net force per unit volume, and the rst two terms on the right are our pressure
and potential gradients. Already, if these terms on the right are unbalanced (e.g., if we have a
spatially-varying density) we will have a net acceleration of the uid, and hence motion. What
ix
Strictly, for a uid of constant density, a spatially-varying density in Eq. 15.21 implies that the velocity eld
must have zero divergence, or be zero everywhere to have a density which does not vary in time. Only the case v =0
corresponds to a truly static situation, and thus, if has any spatial variation, a time variation is implied.
does the acceleration term look like?
What we really need to nd is v /t for innitesimal t, that is our acceleration. Just from the
mathematics of partial derivatives, we can say quite a lot already. Say we know the velocity of a
innitesimal volume of uid at some particular point in space and time, v (x, y, z, t). What is the
velocity of the same bit of uid at some later time t + t when the bit of uid is at a neighboring
point (x + x, y + y, z + z)? From the denition of partial derivatives, for small changes in x,
y, z, and t (i.e., to rst order) we can write the change in velocity as
v = v (x + x, y + y, z + z, t + t) v (x, y, z, t) (15.30)
=
v
x
x +
v
y
y +
v
z
z +
v
t
t (15.31)
This is not incredibly useful, as such, but we can multiply and divide every spatial derivative by
t to put this in a more interesting form:
v =
v
x
x +
v
y
y +
v
z
z +
v
t
t (15.32)
=
v
x
x
t
t +
v
y
y
t
t +
v
z
z
t
t +
v
t
t (15.33)
=
v
x
v
x
t +
v
y
v
y
t +
v
z
v
z
t +
v
t
t (15.34)
=
_
v
x
v
x
+
v
y
v
y
+
v
z
v
z
+
v
t
_
t (15.35)
The acceleration d v /dt then becomes, in the limit t 0,
d v
dt
=
v
x
v
x
+
v
y
v
y
+
v
z
v
z
+
v
t
(15.36)
This might not look like much, but if we look and rearrange it carefully we can recognize a nicer
vector form:
d v
dt
= v
x
v
x
+ v
y
v
y
+ v
z
v
z
+
v
t
(15.37)
=
_
(v
x
x + v
y
y + v
z
z)
_

x
x +

y
y +

z
z
__
v +
v
t
= ( v ) v +
v
t
(15.38)
Can you see why it must be ( v ) v and not, e.g., v ( v )? (If for no other reason, the former
is a vector while the latter is a scalar!)
Having found the acceleration, in the absence of viscous forces our equation of motion is complete:
d v
dt
= ( v ) v +
v
t
= P (equation of motion, no viscosity) (15.39)
We can add a bit more physical content to our equation of motion by dening a new eld from the
curl of the velocity,

= v . This quantity is called the vorticity of the uid, and it characterizes
the circulation of the uid. If

=0 everywhere, the uid is said to be irrotational. By introducing
the vorticity, we can separate the terms in our equation of motion to characterize two basic cases:
uids that swirl, and those that do not.
x
If we are only interested in uids that do not circulate,
this will allow considerable simplication.
In order to achieve this separation, we can also make use of the following vector identity to introduce
terms that contain v :
( v ) v = ( v ) v +
1
2
( v v ) =

v +
1
2
v
2
(15.40)
This allows us to put our equation of motion in the following form:
v
t
+
v +
1
2
v
2
= P (15.41)
Now, taking advantage of this new form, we can consider only irrotational uids for which

=0
(as is the case in our experiment), in which case we have the simpler result
v
t
+
1
2
v
2
= P (equation of motion, no viscosity, irrotational) (15.42)
Finally, there is one more simplication we can make for many reasonable cases: the assumption
of steady ow. This doesnt mean we have nothing happening, it is merely the condition that we
have motion of the uid at constant velocity, v /t =0. In this case,
1
2
v
2
= P (equation of motion, no viscosity, irrotational, steady ow) (15.43)
Since every term in this equation involves a gradient, we may simply integrate both sides to get rid
of a gradient from every term, and once we remember to add in an integration constant, we have
1
2
v
2
+ P + = (const) (15.44)
This is Bernoullis theorem, which is basically a statement of conservation of energy per unit volume
for the uid. Compare this to our starting point, P + =(const.), and you will see that the new
term
1
2
v
2
is nothing more than the kinetic energy of the moving uid!
x
It might help to recall the fundamental theorem of vector calculus, which roughly states that we can build any
reasonable vector eld out of the sum of an irrotational (zero curl) eld and a solenoidal (zero divergence) eld.
Viscosity in one dimension
Adding a viscous (drag) force to our equation of motion is not much of a problem, in principle. If
we have a viscous force

f
v
per unit volume, then Newtons law yields
( v ) v +
v
t
= P +

f
v
(15.45)
The problem is, what is the viscous force? The model of uid ow above basically ignores the pres-
ence of any lateral shear forces, or forces perpendicular to the direction of the uid ow. In other
words, our rst model assumes that the uid will put up no resistance to being pushed around,
which is clearly unrealistic. This is not even realistic for a solid: when we deform a solid, we know
that it will produce a restoring force proportional to the strain it experiences, giving rise to Hookes
law macroscopically. Real uids will also react to an applied force or pressure, but more important
in this case than the amount of strain is the rate at which strain is produced. For example, in most
uids it is easier to move slowly than it is to move rapidly think about swimming or stirring a
jar of thick syrup.
Our previous model ignored any interactions between a moving uid and a solid surface it encoun-
ters. In fact, it would have all but impossible to do so without some sort of empirical guidance
or at least a hint at the answer. In this, we are lucky, however. One important experimental fact
severely constrains models of viscous forces: the velocity of a uid is exactly zero at the boundary
of a solid surface. This is not an obvious fact, but one you can easily verify: how else would your
fan blades have dust on them? Shouldnt it blow o?
With this fact, we can attempt a model of viscous forces. Image that we have two at parallel plates
of area A immersed in an initially stationary uid, separated by a distance d. We hold one plate
at rest in the uid, and move the other plate at velocity v
o
through the uid. In a uid without
viscosity, the moving plate would not disturb the uid at all, and the uid velocity would be zero
everywhere. However, if the relative velocity of uid and plate must be zero at each plates surface,
that means that the uid velocity varies from 0 to v
o
moving from the stationary to the moving
plate! At the surface of the moving plate, the uid must have velocity v
o
, and at the surface of the
stationary plate, it must have v =0.
If you measure the force per unit area required to keep the top plate moving, it turns out to be
proportional to the velocity of the plate divided by the spacing between the plates.
F
A
=
v
o
d
(15.46)
The constant of proportionality is known as the coecient of viscosity, and to some extent it is
a measure of how much force must be supplied to produce motion in a uid. Noting that power is
F v , you can see that the power required to maintain a speed v in a uid scales as v
2
.
d
v = 0
v
F
v
o
Figure 15.3: Viscous drag between two parallel plates in a uid.
In moving beyond a single dimension, we have even more basic problems to consider: if we press
on a uid in one direction, it will move in all directions, not just along the direction of the applied
force. This is in sharp contrast to our usual considerations of innitesimal particles, or rigid objects.
What do we do when the object can squish? In the example above, the moving plate will displace
the uid it is moving through, imparting velocity in the directions perpendicular to the motion of
the plate. Evidently, what we lack is a way to relate displacement along one direction with force
along another. The mathematical tool we are missing is the tensor, a generalization of vectors and
scalars which turns out to be indispensable for many areas of physics.
The Stress Tensor
Tensors
So far as we need them, a tensor is a set of numbers (or a matrix) that when multiplied by a
vector gives back a new vector. Of course, this much can be accomplished by vector or scalar
multiplication, but what makes tensors special is that the two vectors need not be simply parallel
or perpendicular. This is exactly what we need to understand stress and pressure in materials:
relating a displacement or force in one direction to a resulting force along a dierent direction,
particularly when materials are allowed to deform. A close analogy to the type of mathematical
object we need is the rotation matrix: a multiplying a given vector by a rotation matrix gives a
new vector of the same length, but pointing in a new direction. A tensor is a more general type of
matrix, in which both the length and direction of the resulting vector are generally dierent.
As an example, lets say we want to consider the conductivity of a material, . Ohms law states
that the current density is proportional to the electric eld via the conductivity:
j =
E (15.47)
In an isotropic, homogeneous material, is just a scalar, a plain number, that characterizes how
much current density results from a specic electric eld. As such, a scalar conductivity results
in a current density which is always parallel to the electric eld. In many materials, this is a
perfectly reasonable assumption. However, this is clearly a simplication: what about crystals? In
a perfect crystal, we have a symmetric arrangement of atoms which is clearly not isotropic, and it
is unphysical to expect the conductivity to be the same along every direction in the crystal. If we
have a simple cubic crystal, it would be reasonable to expect the conductivity to be the same along
all three crystallographic directions, but we would certainly expect a dierent conductivity along
other directions.
Consider a simple two-dimensional crystal, with a square grid of atoms along the x and y directions.
If the spacing of atoms along x and y is the same, we expect that a given electric eld applied along
the x or y direction would lead to the same current density. However, if we applied the electric
eld along the line y = x, 45
with respect to the rows of atoms, we should expect a dierent

current density. Thus, at the very least, our conductivity must be direction-dependent so long as
the crystal is not isotropic! Moreover, this means that we cant even reasonably expect that the
current density is along the same direction as the electric eld. If the eld is along the the line
y = x, and we have dierent conductivities in the x and y directions, we should expect that the
resulting current density has both x and y components, and they will not be the same. Even in an
isotropic material we have to worry about this to an extent, current will spread out in all directions
in a uniform conductor.
In general, the conductivity actually has nine components relating electric eld to current density,
since we have three directions for E combined with three components for j. The conductivity, then,
is really a matrix:
j
i
=

j
ij
E
j
or
_
_
j
x
j
y
j
z
_
_
=
_
xx

xy

xz
yx

yy

yz
zx

zy

zz
_
_
_
_
E
x
E
y
E
z
_
_
(15.48)
The nine components of make a tensor, relating

E along an arbitrary direction to a resulting
j along a dierent direction. The indices of signify which component of

E is being related to
which component of

j : the rst index is the component of

E , the second the component of

j .
Incidentally, the fact that we require two indices to tabulate all of the components of makes it a
second rank tensor.
xi
Thus, for the x component of

j , we have
j
x
=
xx
E
x
+
xy
E
y
+
xz
E
z
(15.49)
On the other hand, if we apply an electric eld along the x direction only,

j has components in all
xi
By the same logic, we can call vectors rst rank tensors, needing only one index, and scalars zero rank
tensors.
three directions:
j
x
=
xx
E
x
j
y
=
yx
E
x
j
z
=
zx
E
x
(15.50)
Usually, we dont need to deal with all nine components, and we can make use of symmetry to
reduce the number of independent components. For instance, the conductivity tensor is symmetric,
meaning that
ij
=
ji
. In fact, it is possible to simplify the conductivity even further. Our choice of
axes along which to decompose the electric eld and conductivity vectors, and thus the conductivity
tensor, was completely arbitrary. In fact, it is always possible to choose axes such that the tensor
is diagonal, e.g., such that only
xx
,
yy
, and
zz
are non-zero:
_
_
j
x
j
y
j
z
_
_
=
_
xx
0 0
0
yy
0
0 0
zz
_
_
_
_
E
x
E
y
E
z
_
_
(15.51)
In a crystal, nding the diagonal representation of a tensor typically corresponds to choosing the
natural crystallographic axes for decomposing the electric eld and current density. Finally, in an
isotropic material, in which the conductivity is independent of direction,
xx
=
yy
=
zz
, and we
may treat the conductivity as a simple scalar.
Other examples of tensors
In fact, youve already encountered tensors many times, probably without knowing it. Generally
speaking, if you need to relate two vectors, and they in general need not be strictly parallel or
perpendicular, a tensor is probably involved. For instances, the moment of inertia is really a 2nd-
rank tensor, since angular momentum and angular velocity are not in general parallel. Torque is
also a 2nd-rank tensor, and anti-symmetric (
ij
=
ji
), but happens to transform like a vector in
three dimensions. For that reason, we usually just treat it as a vector (or pseudovector, really)
since we can get away with it!
Stress
So what is stress? Essentially, it is nothing more than a generalization of pressure, a net force per
unit area. Hydrostatc pressure we are used to dealing with is just a special type of stress, when
the net force is normal to area of consideration. In a static uid, the force on each side of an
innitesimal cube of uid is the same in magnitude and always normal to the surfaces of the cube.
In this case, the stress is just the hydrostatic pressure, and it is a simple scalar:

F =PA n, where
n is a unit vector normal to the area A.
When we wish to deal with the internal forces in continuous objects, however, this need not be
true. Inside a solid object or a uid, we know there are internal forces between neighboring parts of
the material holding it together. Consider rst a cube of a nice squishy substance like gelatin, and
cut it into two pieces. Clearly, before we cut the gelatin, there must have been a force holding the
two pieces together. Before the cut, each half exerted a force F on the other to hold the block of
gelatin together, so the stress in the material was simply this force divided by the area of the cut
surface. However, the net force between the two pieces was not simply perpendicular to the cut
surface. If that were true, any innitesimal force along the cut plane would have separated the two
pieces. Thus, there must be forces acting not just normal to any surface in the block, but also along
the two tangential directions. In order to properly treat a patch of surface within a continuous
object, we must deal with all three components of force acting on the surface. This is what stress
is, a generalization of pressure to encompass forces acting on a surface in all three directions.
Let us go back to the example above, where we have a at plate moving at velocity v
o
through
a uid. In that case, we had two types of forces present. First, we had a force per unit area on
the surfaces of the plate due to the hydrostatic pressure of the uid, which acted equally in all
directions and normal to each surface. This force can be described by a simple scalar, the pressure,
and the area of the plate. Second, we had a force acting antiparallel to the velocity due to the
viscous drag of the uid. This is what we would call a shear force, being tangential to the surface
of the plate. The force per unit area due to viscous drag is thus a shear stress, acting in the x
direction, and it depends on a velocity in the x direction and an area in the xy plane. A com-
plete description of such forces will require a tensor, the stress tensor. As another quick example,
lets go back to our cube of gelatin. Say we press down on the upper face lying in the xy plane.
This will clearly lead to a net force in the z direction on both faces in the xy plane, and a net
shortening of the cube along the z direction. If the gelatin is incompressible, however, conserva-
tion of matter requires that the cube bulge out in the x and y direction, meaning there must be
outward forces on the other four faces of the cube! Again, we will need a tensor to describe this
situation, since we have an applied force in one direction leading to net forces in all three directions.
How can we gure out what the stress tensor looks like? Lets consider a volume of continuous
incompressible material, of constant density . Now, take a small slice of this material perpendicular
to the x axis, making a little square of sides y and z with area normal x. If we apply a force
F
1
to this surface along an arbitrary direction, we can break it up into components F
1x
normal
to the surface and F
1y
and F
1z
tangential to the surface. The components of stress are just
these forces divided by the area of our surface, labeled with two indices: the rst labeling the
direction of the force component, the second the area normal. For example, the force per unit area
along the y direction is just
yx
=
F
1y
yz
=
F
1y
a
x
(15.52)
where a
x
is just the area of our element of surface perpendicular to the x direction. Similarly, we
have net forces per unit area in the x and z directions,
zx
=
F
1z
yz

xx
=
F
1x
yz
(15.53)
The stress
xx
acts normal to our little area, just as a simply hydrostatic pressure would, while the
stresses
yx
and
zx
act along the transverse directions. In total, just to describe the stress along
a single axis, we need three components, which means that a full description of all the stresses
on an object will require nine components. For example, if we now take a slice of material lying
perpendicular to the y axis, lying in the xz plane, this area will have a net force
F
2
acting on
it, and resolving it along the three axes leads to stresses
xy
,
yy
, and
zy
. We can make a similar
construction for a slice perpendicular to the z axis, and in total the stress on our object will be
characterized by nine numbers, which we can conveniently express as a matrix:
ij
=
_
xx

xy

xz
yx

yy

yz
zx

zy

zz
_
_
(15.54)
The diagonal components
ii
are normal stresses, representing forces per unit area acting perpen-
dicular to the area of a given face. These components are what we would usually just call pressure,
the net force per unit area acting perpendicular to a given face. The o-diagonal components
are the shear stresses acting along the two directions tangential to a given face, analogous to the
tangential frictional force present when we slide two objects past one another. Our nine numbers
ij
in total make up the stress tensor, where i indicates the direction along which the stress acts,
and j indicates the surface normal of the relevant face. Thus,
xy
represents a shear stress acting
in the x direction on a face whose area normal points in the y direction.
At this point, it is probably useful to draw a little picture. Take a small cube of material, aligned
along the x, y, and z axes. Looking down the z axis at one side of the cube, the components of the
stress in the x and y directions acting on four of the faces looks like this:
xx
xx
yx
yx
yy
yy
xy
xy
Figure 15.4: The forces in the x and y directions on the faces of an innitesimal cube. The diagonal components of stress
ii
act normal to each face, while the o-diagonal components
ij
act tangentially to each face. Since the cube is very small,
the stresses do not change appreciably across the cube.
As you can see, the forces on real continuous objects are rather complicated. From our initial
discussion of a static uid, requiring only a simple scalar pressure, we now have a nine-component
second-rank tensor. However, it is not as bad as it seems: the stress tensor turns out to be symmet-
ric, and we dont need all nine components. If we consider an innitesimally small cube of material,
then we can imagine that the stresses do not change appreciable from one side to the other. As
shown in the gure above, the forces on opposing sides of the cube must be equal and opposite in
this case. This also implies that the torque about the center of the cube is zero if it were not,
the cube would start spinning, which would be unphysical for an innitesimally small object. If
our cube has sides of dimension a, then we can easily write down the torque about the center as
a(
yx

xy
) =0, which means
xy
=
yx
. We can apply the same logic looking at the other faces
of the cube, and just by considering that the cube must be in rotational equilibrium we nd that
the stress tensor must be symmetric,
ij
=
ji
. In short, we have only have six unique components
of stress, rather than nine.
Since our stress tensor is symmetric, it can be described by a symmetric matrix. If you have taken
linear algebra, you might recall that this leads us to an even more important property of the stress
tensor: since it is symmetric, it is always possible to nd a choice of coordinate axes for which
it is diagonal. That is, if we choose our coordinate axes carefully, it is always possible to nd a
special orientation for which our stress tensor has only the three components
xx
,
yy
, and
zz
. In
a perfect crystalline material, this special choice may correspond to the crystallographic axes, for
instance. However, in general, the stress tensor varies from point to point in a material, meaning
it is actually a tensor eld. Just like we have a scalar eld T(x, y, z) describing the temperature
everywhere in a room, or a vector eld

E (x, y, z) describing the electric eld through all space, our
stress tensor eld describes the components of stress at all points in a material. At every point in
space, the stress tensor gives us nine numbers six unique numbers describing the forces at that
point, and thus a full description of the forces in a body require six functions of position.
Viscosity and stress in three dimensions
After a long detour, we can nally return to our parallel plates moving within a uid from
Sect. 15.4.1. Recall our setup:
To be a bit more concrete, let the x axis be in the direction of the applied force, the y direction
upward, and the z direction out of the page. Using our new tensor machinery, we can write the
force per unit area along the x direction required to keep the top plate moving as a stress:
yx
=
F
x
yz
=
v
o
d
(15.55)
Thus, what we have previously considered is a shear stress acting tangential to the plates due to
a viscous force along the x axis, and empirically it is found to be proportional to the speed of the
upper plate through a scalar coecient we call the viscosity.
d
v = 0
v
F
v
o
Figure 15.5: Viscous drag between two parallel plates in a uid.
As a slightly more general case, we could forget about the plates, and only consider articial surfaces
within the uid itself, moving at dierent velocities. In Fig. 15.6 below, we look at an extended
region within a moving uid, with its faces parallel to the uid ow. In general, the velocity of
the uid will vary along the vertical direction (y), such that at the top of our cell the velocity is
v + v, and at the bottom it is just v. Based on our simpler model above, the net shear force
acting on the cell in the x direction will the dierence between the forces on the top and bottom of
the cell, divided by the area of the plate A. By analogy with the situation with two plates, the
shear stress is then proportional to the dierence in velocity between the top and bottom of the
cell divided by the vertical extent of the cell:
F
x
A
=
v
y
=
v
x
y
=
yx
(uid velocity along x only) (15.56)
This net shear stress acts to the right on the top face, and would tend to either deform our cell
into a parallelogram or lead to a rotation of our cell in the clockwise direction. The only way that
the uid will be irrotational is if v
x
/y = 0, that is, the velocity is constant along the vertical
direction and there is no net force at all. Otherwise, a variation in uid velocity along the vertical
direction leads to a horizontal shear stress.
What if the velocity of the uid isnt strictly parallel to the faces of our cell? Lets say the vertical
component of the velocity varies across the top and bottom faces, with the velocity being higher
on the left side of the cell. This situation would also tend to cause a clockwise rotation of our cell,
meaning that there must also be stress components along the x direction due to the variation of
uid velocity in the x direction (as well as normal components along the y direction). For a general
uid velocity, the horizontal shear stress must then have two components:
yx
=
v
y
x
+
v
x
y
(15.57)
y
A
F
v +v
v
Figure 15.6: A small volume of uid within a ow.
We could nd the other shear components
yz
and
zx
similarly. Note that this equation immedi-
ately satises our symmetry requirement
xy
=
yx
. We can also see from this general expression
that there are only three cases in which there is no shear stress in the uid: either the uid is static
( v =0, the uids velocity varies only along the out-of-plane z direction (v
x
/y =v
y
/x=0), or
the uid is uniformly rotating (v
x
/y =v
y
/x). Of course, there are also no shear forces in a
uid with zero viscosity, but such things are incredibly rare.
xii
Along these same lines, we can also nd the normal stresses, those acting perpendicularly to the
faces of our cell. For example, if there is a variation in velocity along the vertical direction, then
there will also be a net force on the cell in the vertical direction, along with the shear component:
yy
= 2
v
x
x
(15.58)
The normal components of the stress,
ii
, are what we would simply call pressure if the uid were
static. In the case of a moving uid, the total normal force per unit area would be the static
pressure P plus the normal stress due to the uid motion.
In general, so long as the uid is incompressible, based on our description above you should be able
to convince yourself that the stress components are given by
ij
=
_
v
i
x
j
+
v
j
x
i
_
(15.59)
xii
Liquid helium is a so-called superuid with zero viscosity, a macroscopic quantum-mechanical eect that can
be observed only at very low temperatures.
Viscous forces in three dimensions
Now that we have the shear stresses in the uid in the presence of viscosity, we can complete our
equation of motion. All we need to do is work backwards to determine the forces on an arbitrary
cell within a moving uid from the stress components. Imagine we have again a small cube of uid,
Fig. 15.7, whose faces are aligned with our coordinate axes, with sides of length x, y, and z.
In addition to any hydrostatic pressure, our cube will have a force on each of its six sides due to the
stress in the moving uid uid, whose velocity we will assume to vary in magnitude and direction.
xx
xx
yx
yx
yy
yy
xy
xy
x x +x
y
y +y
1 2
3
4
Figure 15.7: The stresses in the x and y directions on the faces of a cube cube of uid.
First, we can tabulate all of the forces along a given axis, starting with x. All six faces of the cube
will have a stress component in the x direction: four shear forces, and two normal forces. On face
1, we have a normal stress component
xx
acting over an area yz, and net x component of the
force on face 1 will be the product of stress and area. However, we must be careful: the stress
is really a tensor eld, and it varies with position, so the value of
xx
is dierent for face 1 and
face 2, for instance. Thus, we should explicitly note at which position we are evaluating the stress
components. With that in mind,
F
x1
=
xx
x+x
yz (15.60)
The x component force on face two will be similar and opposite in sign, the only substantial
dierence is that we are evaluating the stress tensor at a dierent position:
F
x2
=
xx
x
yz (15.61)
Faces 3 and 4 will also have forces in the x direction through the shear stress
xy
. Face 3 has area
xz and it is located at vertical position y + y. Face 4 has the same area, but the force is in
the opposite direction and
xy
should be evaluated at a vertical position y.
F
x3
=
xy
y+y
xz (15.62)
F
x4
=
xy
y
xz (15.63)
(15.64)
Finally, faces 5 and 6 (on the front and back of the cube in Fig. 15.7) also have forces along the x
direction through the shear stress
xy
:
F
x5
=
xz
z+z
xy (15.65)
F
x6
=
xz
z
xy (15.66)
(15.67)
All that is required now is to tabulate the net force along the x direction for the whole cube:
F
x
= F
x1
+ F
x2
+ F
x3
+ F
x4
+ F
x5
+ F
x6
(15.68)
=
_
xx
x+x
xx
x
_
yz +
_
xy
y+y
xy
y
_
xz +
_
xz
z+z
xz
z
_
xy (15.69)
=
xx
xz +
xz
xz +
xz
xy (15.70)
As with our equation of motion without viscosity, it is more useful to consider the force per unit
volume along x, which well call f
x
:
f
x
=
F
x
xyz
=

xx
x
+

xy
y
+

xz
z
(15.71)
If we take the limit that the dimensions of our cube become innitesimally small, what we have is
the denition of a partial derivative:
f
x
=

x
x
x
+

x
y
y
+

x
z
z
(15.72)
We can repeat the analysis for the forces along the other directions, and our general expression for
an incompressible uid is
f
i
=
3
j=1
ij
x
j
with
ij
=
_
v
i
x
j
+
v
j
x
i
_
(15.73)
The stresses in a uid depend on the velocity gradients in the uid (or, equivalently, the rate of
change of shear strain), while the forces depend on the stress gradients. There will only be a net
force on a volume of uid if there is a net spatial variation of stress, and there will only be stress if
there is a net spatial variation in velocity. Combining the two relationships above, we can cut out
the middleman and relate the viscous force per unit volume directly to the velocity distribution:
f
i
=
3
j=1
x
j
_
v
i
x
j
+
v
j
x
i
_
(15.74)
If we write out all three components of the force and rearrange the terms, we can recover a much
more compact vector equation. Lets rearrange the sum and see what comes out:
f
i
=
3
j=1
x
j
_
v
i
x
j
+
v
j
x
i
_
=
3
j=1
2
v
i
x
2
j
+

x
i
3
j=1
v
j
x
j
=
2
v
i
+

x
i
v (15.75)
Considering all three components of the force per unit volume, we have a nice vector equation in
the end:
f =
2
v + ( v ) (15.76)
Here we have used the vector Laplacian
2
v , which is just
2
v
x
x +
2
v
y
y +
2
v
z
z. We can
make this still simpler, however, by remembering that for an incompressible uid, the continuity
equation reads v =0. With that in mind, after much pain we ultimately have a simple form for
the viscous force in an incompressible uid
f
v
=
2
v (viscous force, incompressible uid) (15.77)
Our pervasive assumption of an incompressible uid does come at a price. For instance, we will
not be able to treat density variations in the uid, such as sound waves. However, a wide variety of
interesting uids are essentially incompressible, and the form of the viscous force per unit volume
above is sucient. This assumption serves quite well for, e.g., air owing at low speeds compared
to the speed of sound.
The equation of motion with viscosity for incompressible uids
The general equation of motion we developed was
( v ) v +
v
t
= P +

f
v
(15.78)
where f
v
was our yet-to-be-determined viscous force. For an incompressible uid, using the form
of the viscous force from Eq. 15.77, our equation of motion reads
( v ) v +
v
t
= P +
2
v (15.79)
This non-linear partial dierential equation is the Navier-Stokes equation for ow of Newtonian
incompressible uids. The Navier-Stokes equation is not straightforward to interpret qualitatively,
and famously dicult to solve in even the simplest cases. Another more common form substitutes
the potential per unit mass = g :
_
v
t
+ v v
_
= P + g +
2
v (15.80)
This form is more easily interpreted as a statement of Newtons law: mass () times acceleration
equals the sum of forces, namely pressure (P), viscous forces (
2
v ), and gravity ( g ). Since
we are assuming constant density (incompressible uid), the continuity equation is simply v =0,
which is a statement of conservation of uid volume.
Incidentally, we can also reintroduce our vorticity

= v :
t
+

v +
1
2
v
2
_
= P + g +
2
v (15.81)
That means that for an irrotational uid (
=0), we have
t
+
1
2
v
2
_
= P + g +
2
v (15.82)
The steady-state (/t =0) equation for an irrotational uid reads
1
2
v
2
= P + g +
2
v (15.83)
Finally, if we are interested in the steady-state behavior of uids at ve
Stokes ow around a solid sphere
If we are only interested in slow and steady ow (small Reynolds numbers) of an incompressible
uid, we may neglect the acceleration term in the Navier-Stokes equation
1
2
v
2
:
P +
2
v + g = 0 (15.84)
This is just one step up from our equation of state for a completely static uid, we now retain only
the viscous force
2
v . Further simplication is possible due to the symmetry of the problem. Let
the uid ow be along the z axis with constant velocity V
, with the solid sphere of radius R at

the origin. In this case, by symmetry the uid momentum is clearly independent of (the angle
in the xy plane). In spherical coordinates the Navier-Stokes and continuity equations read
P
r
+
_
2
v
r
2v
r
r
2

2
r
2
v

2
r
2
v
cot
_
+ g
r
= 0 (15.85)
1
r
P
+
_
2
v
+
2
r
2
v
r
r
2
sin
2
_
+ g
= 0 (15.86)
1
r
2
r
_
r
2
v
r
_
+
1
r sin
(v
sin ) = 0 (15.87)
Here we have expanded the and r portions of the gradient operators in spherical coordinates.
Perhaps surprisingly, the stress distribution, pressure distribution, and velocity components can be
found analytically:
r
=
3
2
V
R
_
R
r
_
4
sin (15.88)
P = P
o
gz
3
2
V
R
_
R
r
_
2
cos (15.89)
v
r
= V
_
1
3
2
_
R
r
_
+
1
2
_
R
r
_
3
_
cos (15.90)
v
= V
_
1
3
4
_
R
r
_
1
4
_
R
r
_
3
_
sin (15.91)
Note the following boundary conditions: at the spheres boudnary r = R, v
r
= v
= 0, and at
r =, v
z
=V
. Equation 15.89 is readily parseable: P

o
is the pressure in the plane z =0 far from
the sphere, gz is the hydrostatic pressure eect, and the term with V
results from uid ow

around the sphere. These equations are valid for Reynolds numbers less than approximately one.
In Fig. 15.8, we show uid ow around a sphere under these conditions.
What we are interested in now is the force on the sphere due to this ow. The normal force (along
the z axis) acting on the solid sphere is due to the pressure given by Eq. 15.89 with r = R and
z =Rcos :
P(r = R) = P
o
gRcos
3
2
V
R
cos (15.92)
The net upward force in the z direction due to the pressure dierence on the top and bottom
portions of the sphere is found by multiplying this pressure times the innitesimal bit of surface
area over which it acts, R
2
sin d d and integrating over the surface of the sphere:
F
n
=
2
_
0
_
0
_
P
o
gRcos
3
2
V
R
cos
_
R
2
sin d d (15.93)
F
n
=
4
3
R
3
g + 2RV
(15.94)
We recover two terms for the normal force: the rst is the buoyant force and the second the form
Figure 15.8: Forces on and streamlines around a sphere in Stokes ow. From http://en.wikipedia.org/wiki/File:Stokes_
sphere.svg.
drag. At each point on the surface, there is also a shear stress acting tangentially,
r
. This
tangential force, since we are dealing with a curved surface, has both x y and z components.
Over the whole sphere, the former will vanish by symmetry, but the latter will give rise to a net
force for any non-zero uid velocity. On any innitesimal patch of surface, the z-component of
this tangential force is (
r
) (sin ) R
2
sin d d, and once again integrating over the spheres
surface we nd
F
t
=
2
_
0
_
0
(
r
[
r=R
sin ) R
2
sin d d (15.95)
From Eq 15.88,
r=R
=
3
2
V
R
sin (15.96)
which results in a net frictional drag from the tangential ow of
F
t
= 4RV
(15.97)
Thus, the total force on our sphere in the owing uid is
F =
4
3
R
3
g + 6RV
(15.98)
The force has two terms, as expected: the rst due to gravity (the weight of the uid), exerted
even if the uid is stationary, and the second associated with uid motion, sometimes called the
drag force. Both forces act in the same direction, opposing the direction of uid ow.
Equation 15.98 is known as Stokes law, and from it we may determine the terminal velocity of
a falling sphere. Consider a sphere falling in a stagnant uid of density
s
. In this case V
is
the relative velocity of the uid with respect to the sphere, which in this case is just the velocity
of the falling sphere since the uid is stationary. The static and drag Stokes forces act opposite
the direction that the sphere falls, and at the terminal velocity V
t
, precisely balance the spheres
weight. If the sphere has density
s
, this means
4
3
R
3
g + 6RV
t
=
4
3
R
3
s
g (15.99)
This leads to a terminal velocity of
V
t
=
2gR
2
(
s
)
9
(15.100)
As expected, it is the relative density of uid and solid that determine the behavior of the sphere:
if the sphere is more dense than the uid, it sinks, and if it is less dense than the uid, it rises.
In the present experiment, the uid density (air) is negligible compared to the solid density (oil),
and we arrive at Eq. 15.5.
Millikans correction to Stokes ow for low velocities
16
Gamma ray spectroscopy
Note: It is assumed that you have read the introduction to the previous experiment, gamma ray
attenuation. See Sec. 6.
16.1 Introduction
The purpose of this lab is to become familiar with the response of a NaI detector to gamma rays and
to identify unknown gamma sources. NaI, doped with thallium, is an inorganic scintillator detector,
that is, it produces scintillation light when excited by the ionizing radiation of charged particles.
Gamma sources are unstable nuclei which emit photons when they decay. The combination of
gamma rays emitted by an unstable nucleus is unique. Besides gamma rays emitted due to nuclear
de-excitations, annihilation radiation may be produced if the decay products include a positron:
the positron annihilates with an atomic electron to produce two 511 keV photons.
i
16.1.1 Scintillation Counters
When radiation interacts with certain substances, a small ash of visible light (a scintillation) is
produced. This ash of light can be detected by a special kind of photocell (called a photomultiplier
tube) which is capable of multiplying the resulting signal by a very large factor. The combination
of a scintillator mounted on a photomultiplier tube is referred to as a scintillation counter.
Depending on the type of scintillator, these counters are capable of detecting as well as determining
the energy of alpha, beta, gamma, and x-rays. To detect alphas, zinc sulphide and plastic phosphors
may be used, while betas are usually best detected by anthracene or plastics. The best scintillator
for gammas (the type of radiation studied in this experiment) is a large crystal of sodium iodide
containing a trace of thallium as an activant. This is referred to as a NaI(Tl) crystal. The crystal
used in this experiment is roughly 4 cm in diameter and 2.5 cm thick, and it is mounted in an alu-
minum can which is optically coupled to the end of the photomultiplier tube. The entire detector
assembly is contained within the aluminum can to shield it from external light.
The scintillation counter has several advantages over the Geiger-Mller counter for the detection
of gamma rays. These advantages include:
greater detection eciency, due to the use of a solid (rather than a gaseous) detector
capability for handling higher counting rates (due to shorter resolving times)
i
Much of this procedure is reproduced from ref. [4]. This version is only for use in PH255 at the University of
Alabama.
179
capability for energy analysis (due to the fact that the amplitude of the output pulse is
proportional to the energy absorbed in the scintillator)
16.1.2 Compton Scattering
The energy of gamma rays is transferred to the NaI(Tl) scintillator chiey by two processes: the
photoelectric eect and the Compton eect. At higher energies (>1.022 MeV), pair production also
plays a role, and will be discussed further below. In the photoelectric eect, the entire energy of
the gamma ray is transferred to the phosphor, resulting in what is known as a photopeak or full
absorption peak. Put another way, the photons in the incident radiation are of high enough energy
to release electrons from atoms and molecules in the detector crystal. Photons in the beam are
thus annihilated, leading to a corresponding reduction in intensity. This photoeect will lead to a
peak in the energy spectrum, known as the photopeak or full absorption peak.
In a real experiment, however, you will notice more structure in your spectrum than just photopeaks
corresponding to gamma emission. Specically, we must also consider the scattering of gamma pho-
tons by electrons in the detector, i.e., the Compton eect. This will lead to two additional features:
a Compton edge and a backscatter peak. Recall that the Compton eect is a collision of
a photon with an electron in which relativistic energy and momentum must be conserved. After
the collision, the photon imparts some of its energy and momentum to an electron, and thus the
photon exits the collision with a slightly lower energy and a new direction. This process is shown
schematically in Fig. 16.1.
E
e
= hk
p
= hk
p
e
Figure 16.1: Illustration of Compton scattering. An incident photon of energy E and momentum hk scatters o of an
electron at rest. The photon emerges at angle with reduced energy E
and momentum hk
. The electron is ejected with

energy E
e
and momentum p
e
.
Analyzing the collision and conserving relativistic energy and momentum, one can readily determine
the exiting photons energy as a function of its energy and ejection angle:
E
=
mc
2
(1 cos ) + mc
2
/E
(16.1)
Here mc
2
is the electron rest mass, approximately 511 keV. This formula might appear more familiar
in terms of the incident and exiting photon wavelengths:
= +
h
mc
(1 cos ) (16.2)
In short, a gamma enters the detector crystal and Compton scatters o of an electron which is
then detected by scintillation. The Compton-scattered gamma leaves the detector, and the amount
of detected energy is thus the kinetic energy given to the electron, or the dierence between the
incident and exiting photon energies. Compton scattering necessarily results in a small energy
transfer than the photoelectric process, where the entire incident photon energy is transferred to
an electron. The maximum possible energy transfer denes the so-called Compton edge," and
occurs when the gamma is scattered backwards at =180
. This maximum energy transfer is thus:

E
ce
= E
(=180
) =
2E
2
mc
2
+ 2E
(16.3)
The Compton edge represents the maximum energy transfer to an electron, or the maximum energy
loss by the photon. Of course, the collision may be considerably more gentle and result in a smaller
energy transfer. Considering all possible collisions, one arrives at a broad distribution of events at
energies less than the Compton edge, and a cuto at the edge itself. For a 662 keV
137
Cs gamma,
the Compton edge occurs at approximately 478 keV.
Still one more eect of Compton scattering must be considered. Gammas may also interact with
electrons outside the detection region, and these electrons may be Compton-scattered back into the
detector and be detected by the photoeect. Only a few angles near 180
will result in electrons

entering the detector, which results in a peak in the energy spectrum at an energy corresponding
to that of the exiting photon:
E
bsc
= E
E
ce
= E
(=180
) =
E
o
1 + 2E
o
/mc
2
(16.4)
Essentially, the backscatter peak arises from absorption in the detector of scattered gammas from
Compton interactions in the material surrounding the scintillator, mainly in the glass layer between
the crystal and the cathode of the photomultiplier. For a 662 keV
137
Cs gamma, the backscatter
peak should occur at approximately 184 keV. In Fig. 16.2 we show what the gamma ray spectrum
from
137
Cs might look like as measured with an ideal detector (in which backscattering does not
occur). At the lowest energies, an x-ray emission from the
137
Ba daughter is visible, followed by
the Compton spectra. The Compton spectrum increases monotonically until the Compton edge is
reached, after which the photopeak is observed. The real
137
Cs spectrum, however, shows several
other features, most notably the backscattering peak.
In the analysis of complex scintillation spectra, care must be taken to identify backscatter peaks and
Compton electron distributions to avoid confusing this with photoelectron peaks. It is also notable
C
o
u
n
t
i
n
g

r
a
t
e

Energy
137
Ba x-ray
Compton
spectrum
Compton
edge
photopeak
0 200 400 600 800
10
100
1k
10k

137
Cs
I
n
t
e
n
s
i
t
y

(
c
o
u
n
t
s
)
E (keV)
660 keV
Photopeak
478 keV
Compton edge
184 keV
Backscatter
~30, 70-90 keV
Cs, Pb X-rays
Figure 16.2: Left: Gamma spectrum of
137
Cs for an ideal detector and no backscattering. Right: Measured gamma spectrum
of
137
Cs. Aside from the main photopeak at 660 keV, clear Compton edge and backscatter oeaj are visible, as well as X-ray
absorption edges. Note the logarithmic scale on the vertical axis.[3]
that the photopeak obtained with a real detector is broadened compared to the ideal case. The
amount of broadening depends on the properties of the particular scintillator and photomultiplier
used. A quantitative measure of the broadening is the resolution, which is calculated as the ratio
of the width of the photopeak at one half its maximum intensity (the full width at half-maximum,
or FWHM) divided by the pulse height (or energy where the maximum occurs, expressed as a
percentage.
16.1.3 Pair Production
The antiparticle of the electron (e
or
) is the positron (e
+
or
+
). When a positron and electron
collide, they annihilate each other and produce two photons, each of energy corresponding to their
rest energy of 511 keV and traveling in opposite directions to conserve linear momentum. Some
radioactive isotopes, particularly those with proton-rich nuclei, decay by positron emission. This
reduces their nuclear charge and atomic number by one. The 511 keV photons (gammas) from the
subsequent annihilation of the positron can be observed in gamma ray spectroscopy, clear evidence
of nuclear annihilation.
A positron can also be produced along with an electron as a pair by a gamma ray with energy
greater than two electron masses (1.022 MeV). That is, the photon be annihilated into an electron-
positron pair. This process conserves charge as well as linear and angular momentum, and is usually
referred to as pair production. Any excess energy above 1.022 MeV is found as kinetic energies of
the pair. After pair production, the positron will be annihilated by another electron, producing
two 511 keV gammas. This is shown schematically in Fig. 16.3.
Pair production and positron emission have a number of eects on gamma ray spectra. Annihilation
511 keV
511 keV
E
> 1.022 MeV

e
+
(
+
)
e
)
Figure 16.3: Illustration of pair production. A gamma enters the detector from the left with energy E >1.022 MeV, producing
an electron-positron pair in the detector. The positron is annihilated producing two 511 keV gammas.
will produce two 511 keV gammas moving in opposite directions, only one of which may possibly
enter your detector if the annihilation occurs outside the detector. Positrons can only travel very
short distance before being annihilated, meaning annihilation almost always occurs in the source
material. The result is a 511 keV photopeak and Compton distribution in the gamma spectrum, as
shown in Fig. 16.4.
Gammas with energies greater than 1.022 MeV can enter your detector and interact by pair pro-
duction, producing a positron inside the detector, as shown in Fig. 16.3. Annihilation occurs so
rapidly that the light produced by the two 511 keV gammas combines with the light produced from
the total kinetic energies of the electron and positron to produce a pulse that represents the energy
of the original gamma. This appears as a photopeak pulse. Sometimes one or both of the 511 keV
gammas escape the detector, and this creates photopeaks that are missing 511 keV of energy (rst
escape peak) or 1.022 MeV (second escape peak).
0 200 400 600 800 1000 1200 1400
100
1k
10k
1274 keV
511 keV
annihilation
peak

22
Na
I
n
t
e
n
s
i
t
y

(
c
o
u
n
t
s
)
E (keV)
Figure 16.4: Gamma spectra of
22
Na, showing a 511 keV annihilation peak..
Two other mechanisms are worth noting, though they do not change the essential physics of the
184 16.2 Objective
problem.
ii
A fourth mechanism, Rayleigh scattering, corresponds to the elastic (photon-energy-
conserving) scattering of light, which simply changes the direction of incident photons. Since no
energy is lost, this process will merely reduce the measured intensity of radiation (because some
photons will be deected out of the detection region) and will not alter the spectrum.
Finally, in principle must also consider direct absorption of incident photons by the nucleus itself,
photonuclear absorption. This process usually results in the ejection of one or more neutrons and/or
protons. This interaction can contribute 510% to the total photon interaction, though within
a fairly narrow energy region usually occurring somewhere between 5 MeV and 40 MeV. At the
energies of the present experiment (<2 MeV), this eect is negligible.
16.2 Objective
The objective of this experiment is to apply understand the interaction of high-energy gamma-ray
photons with matter. You will perform spectroscopic measurements on known materials to identify
and explain the key features in the spectra, and apply your knowledge to the determination of an
unknown material by analysis of its gamma spectrum. You will also be able to determine the rest
mass-energy of electrons.
Hypothesis: Gamma ray spectra are characteristic of particular nuclei, and may be used as an
analytical tool. The spectra can be explained by considering the photoelectric eect, Compton
scattering, pair production, and characteristic x-ray emissions.
1. How does a photomultiplier tube work?
2.
Taylor[1], Ch.
Pfeer & Nir[2], Ch.
http://en.wikipedia.org/wiki/Gamma_ray
See [1719] for relevant nuclear data and [20] for example gamma spectra.
ii
Of course, there are still other mechanisms, such as nuclear-resonance scattering and Delbrck scattering, but
they are negligible from our point of view.
16.4 Supplies 185
16.4 Supplies
2. NaI(Tl) detector
4. Pb cylinder and brick for shielding
5. radioactive sources
Pb brick
Pb cylinder
detector
detector
housing
sources
control electronics
calipers
Pb, Al, Cu
sheets
Figure 16.5: The Spectech UCS30 spectrometer and accessories.
16.5.1 Startup
2. run the program UCS30 from the shortcut on the desktop
3. place the
137
Cs source on top of the lead brick, and place the lead cylinder around it
4. place the cylindrical NaI detector inside the lead cylinder
will take several minutes. The nal spectra should look like the one shown below in Fig. 6.5.
1. Once the calibration is nished,
16.6 Data analysis
For each source, measure the photopeak, the Compton edge, and the backscatter peak energies.
The Compton edge is at an energy corresponding to about half the change in the count rate from the
trough between the photopeak and edge to the immediate maximum in the edge. The photopeak
should be easily identied, and the backscatter peak should be more triangular and asymmetric in
shape, with the peak at or slightly rounded. The energy of the backscatter peak corresponds to
the lowest energy of the top of the peak. Test your energy measurements by verifying that
E
= E
ce
+ E
= E
bsc
+ E
ce
(16.5)
Compute the position of the Compton edges and backscatter peaks for each gamma and indicate
their ideal locations on the spectra.
Solve Eq. 16.3 for mc
2
as a function of E
and E
ce
. Using the photopeak and Compton edge en-
ergies, calculate the mass energy of the electron and the measurement error in that mass. Average
your results to obtain one value for the electrons rest mass energy and the standard deviation in
that value.
Compute the percent resolution for each peak.
The resolution should improve (get smaller) as the energy increases. Is that what you observe?
Further, if the distribution of channel numbers for a given energy deposit truly follows gaussian
statistics, the resolution is proportional to E
1/2
. How does the expectation of gaussian statistics
compare with your observations?
For each of the unknown sources:
1. Estimate the energies of all prominent peaks. (If one of the peaks has an energy consistent
with 511 keV, you may assume that it is the annihilation peak and thus that the parent emits
positrons.)
2. Identify the parent. Show your reasoning. You may use the on-line table of gamma energies.[17,
18] (Use your uncertainty in the measurement of the peak energy as a guide in setting the
width of the window of gamma ray energies in which you search.) You may assume that the
half-life of the unknown source is at least 1 month.
3. If there are several candidates, look at the list of gamma energies and corresponding branching
ratios for each candidate. Are there any other gammas with a comparable or larger branching
ratio than the ones you see? (The branching ratio is labeled as I
g
(%) in the on-line tables.)
The branching ratio indicates the fraction of decays in which the gamma is emitted, smaller
is less likely.) If so, are peaks present in the spectrum corresponding to these gammas? If
not, the candidate is ruled out.
4. Identify the source of each peak, e.g. gamma from nuclear decay, sum of two gammas from
nuclear decay, sum of a gamma from nuclear decay and annihilation gamma, etc.
reports).
The backscatter peak in each spectrum should change in intensity with the addition of a lead sheet.
Could any other material work as well? It is the product of the electron density and the material
thickness that correlates directly with the intensity of the backscatter peak.
Gamma sources available
Table 16.1: Disk gamma source set
Isotope Activity (Ci) Half-life Peaks (keV)
133
Ba 1 10.8 yrs 81, 276, 303, 356, 384
109
Cd 1 462 days 88
137
Cs 1 30.2 yrs 662
57
Co 1 272 days 122, 136
60
Co 1 5.27 yrs 1173, 1333
54
Mn 1 313 days 835
22
Na 1 2.6 yrs 511, 1275
207
Bi 31.6 yrs 569.7, 894, 1063.66, 1770.2
unknown 1.5 ? ?
Sources from Spectrum Techniques, Oak Ridge, TN (www.spectrumtechniques.com). USNRC
and state license exempt quantities.
17
Charge Transport
17.1 Introduction
In this lab, you will rst learn two dierent but related techniques for measuring electrical resis-
tivity, in this case of a silicon wafer: the linear four-point probe method, and the van der Pauw
method. While the former is more straightforward, the latter is a more general technique appli-
cable to essentially arbitrary samples.
i
In order to learn how to perform high-quality electrical
measurements, you will rst investigate noise characteristics in resistive circuits and the ability of
signal averaging to reduce them.
17.2 Noise in resistive devices
In electronics and communication systems, noise is a random uctuation or variation of an elec-
tromagnetic analog signal such as a voltage or a current. Electronic noise is a characteristic of all
electronic circuits. Depending on the circuit, the noise generated by electronic devices can vary
greatly and arise through several dierent mechanisms. Contributions such as thermal noise and
shot noise are inherent to all devices, while other types depend mostly on manufacturing quality
and defects. Though noise usually has a negative connotation, it does have its uses for instance,
noise power is used in low-temperature thermometry, and the study of noise can be a powerful
technique for elucidating the microscopic mechanisms of conduction.
A noise signal is typically considered as a linear addition to a useful information signal, typied
in Fig. 17.1 where a noise signal is superimposed on constant voltage signal. Noise is a random
process, characterized by stochastic properties such as its variance, distribution, and spectral den-
sity. The spectral distribution of noise can vary with frequency, so its power density is measured in
watts per hertz (W/Hz). Since the power in a resistive element is proportional to the square of the
voltage across it, noise voltage (density) can be described by taking the square root of the noise
power density, resulting in volts per root hertz (V/
Hz). Integrated circuit devices, such as opera-

tional ampliers commonly quote equivalent input noise level in these terms (at room temperature).
Noise levels are usually viewed in opposition to signal levels and so are often seen as part of
a signal-to-noise ratio (SNR). Typical signal quality measures involving noise are signal-to-noise
ratio (SNR or S/N), signal-to-quantization noise ratio (SQNR) in analog-to-digital conversion, and
i
If you are not familiar with electrical measurements at this point, please review Appendix 2 and 3.
189
190 17.2 Noise in resistive devices
compression, peak signal-to-noise ratio (PSNR) in image and video coding, carrier to noise ratio
(CNR) before the detector in carrier-modulated systems, and noise gure in cascaded ampliers.
0 20 40 60 80 100 120 140 160
3.2515
3.2516
3.2517
3.2518
3.2519
3.2520
3.2521
3.2522
3.2523
3.2524
3.2525
3.2526
3.2527
2.5k @ 1.5mA
128 readings/point
256 points

V
o
l
t
a
g
e

(
V
)
time (s)
0 10 20 30 40 50 60

Counts
Figure 17.1: (left): Time-domain signal showing
random noise. (right): Distribution of the noise
about the average voltage, approximately Gaussian.
As with our investigation of counting statistics (Sec. 5), the uctuation of, e.g., a voltage signal
about its mean value is governed by a random process, namely the collisions of the electrons making
up the current. In this regard, the analysis of electrical noise will closely parallel the analysis of
counting statistics in radioactive decay, and both electrical signal and counting noise can be reduced
through signal averaging techniques. Crucial dierences arise, however, due to the diering physical
origins of the uctuations in the two systems. In the case of electrical noise, one may also eectively
reduce uctuations by controlling the measurement temperature and bandwidth and the overall
resistance of the device under test.
17.2.1 Thermal Noise
Thermal noise, or Johnson-Nyquist noise, is the electrical noise generated by random thermal agita-
tion of the charge carriers in an electrical conductor at equilibrium. The direction of the movement
of the charge carriers is changed by collisions with, for example, the host crystal, impurities and
other electrons, resulting in a random movement at zero voltage (Brownian motion), and a certain
degree of randomness in their movement at nite voltages. This random movement gives rise to
uctuations in the net current in the conducting material. These uctuations time-average to zero,
but the time-average of the square of the uctuations is not. Since thermal noise results from a
random process, its amplitude very nearly follows a Gaussian probability density function, as shown
in Fig. 17.1.
ii
Since the movement of electrons is only correlated within the time between two collisions, there is
essentially no frequency-dependence of the noise for frequencies lower than the reciprocal collision
time 1/ of the electrons. The collision time determines the conductivity of a material. In Cu,
for example, 2.5 10
14
s, so below 1/ 40 THz thermal noise is frequency-independent, or
ii
Thermal noise is, formally, distinct from shot noise, a type of electronic noise that occurs when the nite
number of electrons is small enough to give rise to detectable statistical uctuations. While shot noise will dominate
only at very low currents, thermal noise is always present.
17.2 Noise in resistive devices 191
white."
iii
Thermal noise can be modelled as a resistor representing the device under test in series with
a random voltage source, or in parallel with a random current source. This is because by using
Thvenin and Norton equivalents, any linear electrical network consisting of combinations of voltage
sources, current sources, and resistors can be replaced with a single resistor and a series voltage
or parallel current source. Thus, an ideal, noise-free resistive device can be replaced by an ideal,
noise-free resistor in series with a random voltage source. The root mean square noise voltage
produced in a resistor R at temperature T is given by
V
2
) = V
2
rms
= V )
2
+
2
= 4k
B
TRf (17.1)
where k
B
is Boltzmanns constant and f the bandwidth of the measurement. If your measurement
uses ampliers with a lower cuto frequency f
l
and an upper cuto frequency f
h
, the eective
bandwidth of the measurement can be expressed as
f =

2
f
2
h
f
h
+ f
l

2
f
h
(f
h
f
l
) (17.2)
As a concrete example, lets suppose that R=10 k, T =295 K, and our bandwidth is set by lters
of 1 Hz and 100 kHz (f 157 kHz). Then we nd V
rms
16 V. This represents the limit for the
smallest voltage we can resolve across this resistor in this bandwidth. Note that if we put two
resistors in series, the mean square voltage is given by
V
2
rms, tot
= 4k
B
T (R
1
+ R
2
) = V
2
rms, 1
+ V
2
rms, 2
(17.3)
The noise powers add, not the noise voltages. For an arbitrary resistive circuit, we can nd the
equivalent noise by using a Thevenin (Norton) equivalent circuit or by transforming all noise sources
to the output by the appropriate power gain (e.g. voltage squared or current squared).
We can already draw two important conclusions: the noise in a given resistive device can be
eectively reduced by reducing the temperature of the device, or by reducing the measurement
bandwidth. While the former is not always feasible, a number of clever solutions exist for reducing
measurement bandwidth. A simple technique we will use here is signal averaging, which reduces
the bandwidth by repeatedly sampling the quantity of interest. More sophisticated techniques,
such as lock-in detection (modulation) will be explored in future labs.
A third factor inuencing the noise voltage is the resistance of the circuitry under study. Even
in the simplest possible system, a short-circuited resistor, the noise voltage produced increases as
R, and a noise current

_
I
2
) =
_
V
2
)/R =
_
4k
B
Tf/R must be produced. Decreasing the
iii
The frequency spectrum is the Fourier transform of the signal in the time domain. If the signal is random in the
time domain, it can be thought of as containing all possible frequencies superimposed in equal amounts.
192 17.2 Noise in resistive devices
resistance of the circuit under study, if feasible, will reduce the noise in the measured voltage at
the expense of the noise in the measured current. Thermal noise is intrinsic to all resistors and is
not a sign of poor design or manufacture, although resistors may also have excess noise.
In addition to thermal noise in the device under study, one must also consider the noise present
in the components making up the sources, ampliers, and meters used. For example, consider a
simple circuit consisting of a single resistor connected to a current source, and we wish to measure
the voltage produced on the resistor. Independent of the current level, the thermal noise voltage
measured on the resistor will remain the same. However, the current source itself is not perfect, and
the current supplied will have its own variance I
2
). These current uctuations lead to additional
voltage uctuations measured on the resistor. Since the current source uctuations are independent
of the thermal uctuations in the resistor, we must add the variances in quadrature (i.e., add the
noise powers):
V
2
)
tot
= V
2
)
source
+ V
2
)
thermal
(17.4)
17.2.2 Random Telegraph Noise
Random telegraph noise, also called burst, popcorn, impulse, or bi-stable noise, is another type of
electronic noise that is frequently encountered, particularly in semiconductors. This type of noise
consists of sudden step-like transitions between two or more discrete voltage or current levels, as
high as several hundred microvolts, at random and unpredictable times. An example is shown in
Fig. 17.2. Each shift in oset voltage or current often lasts from several milliseconds to seconds, and
sounds like popcorn popping if hooked up to an audio speaker. No single source of popcorn noise
is theorized to explain all occurrences, however the most commonly invoked cause is the random
trapping and release of charge carriers at thin lm interfaces or at defect sites in bulk semiconductor
crystal. For the present experiment, we wish to avoid telegraph noise as much as possible.
0 50 100 150 200
4.231
4.232
4.233
4.234
4.235
4.236
4.237
~2.7

V
o
l
t
a
g
e

(
V
)
time (s)
3.5k @ 1.5mA
128 readings/point
256 points
Figure 17.2: A voltage signal exhibiting telegraph
noise in addition to random noise. The jump in
resistance is 0.08%.
17.3 Signal averaging 193
17.3 Signal averaging
Once the sources of noise in a measurement are understood, the usual question is how they may
be reduced or eliminated. Thermal noise cannot be eliminated, but can be reduced by reducing
device resistance, temperature, or bandwidth. Once these parameters have been controlled, the
remaining noise can be further reduced through signal averaging techniques. In its simplest form,
signal averaging just means the repeated measurement of the quantity of interest, an application
(or exploitation) of the ideas you explored Sec. 5.
As a more concrete example, our goal is to measure the voltage across a resistor. Rather than
making only a single measurement, we may improve our accuracy by taking N measurements and
reporting the mean result V . As we make more and more measurements, the uncertainty in our
mean voltage compared to the true voltage is given by the standard deviation of the mean
s
V
=
s
N
(17.5)
where s is the standard deviation of our collection of measurements. The relative uncertainty of
the mean decreases as 1/
N, and this is at the heart of signal averaging: repeated measurements

lead to increased accuracy, at the price of increased measurement time. The latter point cannot
be overestimated. While the uncertainty in the mean voltage is reduced as O(1/
N), the mea-

surement time increases as O(N), and thus there exists for any measurement, due to practical
considerations, a point at which further reduction of noise through signal averaging is no longer
feasible. For example, to reduce the signal to noise ratio by a factor of two, a factor of four increase
in the number of measurements, and thus measurement time, is required.
Figure 17.3 shows an example of the eect of averaging repeated measurements. In this measure-
ment, a 500 resistor was measured with a 1.5 mA constant current, and the measurement was
repeated up to 1024 times. The ratio of the average voltage to the standard deviation the signal
to noise ratio increases steadily as

N.
17.4 Summary of Noise and Averaging
The root-mean square (RMS) voltage is the sum of the squared average voltage and the squared
standard deviation for a series of N measurements V
i
is
V
2
) =
_
N
i=1
V
2
i
N
= V
2
rms
= V )
2
+
2
= 4k
B
TRf (17.6)
For thermal noise, the RMS voltage is governed by the resistance R and temperature T of the
device under study and the measurement bandwidth f. With no dc component present,
194 17.5 Four point probe techniques
0 10 20 30
2
3
4
5
6
7

1
0
3

<
V
>
/
1/2
Figure 17.3: Eect of point averaging on a volt-
age signal. The relative uncertainty /V decreases
as

N, while the signal-to-noise ratio V / in-
creases as

N.
V
2
) = V
2
rms
= 4k
B
TRf (17.7)
If we drive a resistive device with a current source which also has inherent uctuations, the total
noise voltages from the resistive load and current source add in quadrature. That is, we add the
noise powers:
V
2
)
tot
= V
2
)
source
+ V
2
)
thermal
(17.8)
The signal to noise ratio (SNR) when a dc component is present is dened as the ratio of the
mean voltage to its standard deviation. For a large collection of measurements of size N, the SNR
increases as

N:
SNR =
V
N (17.9)
17.5 Four point probe techniques
Armed with a knowledge of quasi-realistic devices and instruments, we are ready to discuss the
most basic and essential task: measuring the resistivity of a specimen, such as a homogenous wire,
a thin or thick lm, or a bulk sample. Resistivity is of primary interest since it is a geometry-
independent quantity, characteristic of a given material and its processing. However, resistivity
must be determined from a geometry-dependent resistance measurement. Thus, in order to deter-
mine resistivity we must account for the specimen and measurement geometry.
We will assume in the following discussion that the necessary precautions have been taken with
regard to instrumentation, and that all sources and meters can be considered essentially ideal.
iv
iv
Finite wire and contact resistances remain, however, necessitating a four-terminal measurement; see Appendix 3.
17.5 Four point probe techniques 195
Given a homogeneous conducting sample of one (wire), two (lm), or three (thick lm or bulk)
dimensions, a primary quantity of interest is the resistivity (or the conductivity =1/).
17.5.1 Four-point Probe Measurements
Resistivity measurements are commonly performed in a linear four-point geometry using current
sourcing, as shown in Fig. 17.4. In this arrangement, the sample of interest is contacted by four
collinear probes of negligible size compared to any sample dimension. The outer two probes in-
troduce a current I, while the inner two probes measure a potential dierence V . It is clear
that this arrangement will not be subject to artifacts due to nite wire and contact resistances as
discussed in the Appendices. If the voltmeter measuring potential dierence V is nearly ideal
(i.e., the sample resistance between points V
+
and V
is small compared to its internal resistance),

the current drawn by the voltmeter is essentially zero. Thus, the potential dierence measured is
characteristic of the specimen only. From the measured potential dierence and the known source
current, the resistivity of the sample may be determined from by geometric considerations. We
need only determine for various sample geometries how the current ows outward from a source
probe and the resulting potential dierence at the inner probes, superposition and symmetry will
do the rest.
Figure 17.4: Linear four-point-probe measure-
ment. Current is introduced with the outer probes,
and potential dierence is measured between the in-
ner probes.
It is simplest to start by considering a current I introduced into an innite bulk specimen of constant
resistivity through a single probe somewhere in the interior. Charge conservation dictates that
the total current through a sphere of radius r centered on the probe must be constant, and thus
the current density a distance r from the probe must be
J(r) =
I
4r
2
(17.10)
If the conductor obeys Ohms law, we must have
J(r) =
1
E =
1
V (17.11)
The potential (relative to a distant point) due to this current a distance r from the source point
then follows readily since the problem is radially symmetric and the current density is constant at
a given radius r:
J(r) =
I
A(r)
=
I
4r
2
=
1
V
r
(17.12)
V (r) =
I
4r
(17.13)
17.5.2 Bulk samples
Next, we consider a half-innite bulk specimen with one free surface, Fig. 17.5, where the current
is injected through a single probe I
+
. Assuming current still spreads out uniformly, it is now
conned entirely in one hemisphere, doubling the current density at an arbitrary point r and
therefore doubling the potential:
J(r) =
I
2r
2
(17.14)
V (r) =
I
2r
(17.15)
ment on a bulk sample. The outer and inner
probes are separated by a distance a, while the in-
ner probes are separated by b. Current is introduced
by the outer probes, and potential dierence mea-
sured on the inner probes.
In the four-point probe measurement, the potential dierence of interest is measured with probes
V
+
and V
, at distances a and a + b from the source probe, respectively. The potentials at the
probes due to the current injected at I
+
are now easily found:
V
+
=
I
2a
due to I
+
(17.16)
V
=
I
2 (a + b)
due to I
+
(17.17)
Of course, this is not the whole story: we have a second current probe I
, our current sink,"

which we must take into account. By symmetry, the result is the same as above if + and are
interchanged and the sign of the current reversed. The total potential dierence between the voltage
probes V
+
and V
due to a current through both probes I

+
and I
follows by superposition, or
simply doubling the potential dierence due to the single current probe I
+
:
V = V
+
V
=
I
_
1
a

1
a + b
_
=
Ib
a (a + b)
(17.18)
This can be readily inverted to determine the resistivity in terms of the measured potential
dierence and known current and probe spacing:
=
a (a + b) V
bI
(17.19)
For the special (but common) case of equally spaced probes (a = b), we have
=
2aV
I
(17.20)
17.5.3 Thin Films
More relevant for spintronics is the case of a thin lm specimen, Fig. 17.6. Specically, an innite
two-dimensional sheet whose thickness is small compared to the probe spacing, such that we may
approximate the current density as uniform in the direction perpendicular to the lm plane. In
this case, the current no longer spreads evenly in a hemisphere, but is conned within the lms
thickness d. In the plane of the lm, the current spreads evenly leading to circular equipotential
surfaces. The crucial dierence is that at a lateral distance r from a current probe the total current
I must pass through an area dictated by the circumference a circle of radius r and the lm thickness,
and thus the current density is
J(r) =
I
2rd
(17.21)
ment on a thin lm sample of thickness d. The
outer and inner probes are separated by a dis-
tance a, while the inner probes are separated by
b, {a, b} d. Current is introduced by the outer
probes, and potential dierence measured on the
inner probes.
The potential at a distance r is then
V (r) =

2d
ln r (17.22)
Following similar reasoning as above, the potential dierence between the voltage probes is
V =

d
ln
_
a + b
a
_
(17.23)
Which gives the resistivity of the lm as
=
V d
I
1
ln
_
a+b
a
_ (17.24)
With equally spaced probes (b = a),
=
V d
I ln 2
4.53
V
I
(17.25)
We note in passing that for thin lm specimens, it is common to quote a sheet resistivity,
s
= /d,
the resistivity per unit thickness, as well as a sheet resistance R
s
, the resistance per unit thickness.
For equally-spaced probes the, sheet resistivity is
s
= V d/I ln 2.
17.5.4 Wires
Finally, in the one-dimensional case we consider a wire of suciently small cross-sectional area
A that the current may be considered uniform along the radial direction of the wire. That is, we
consider a wire whose radial dimensions are smaller than any of the contact spacings or dimensions,
Fig. 17.7. In this case, the current spreads uniformly throughout the wires cross section, and the
current density is constant throughout, J = I/A. At a distance z from a current probe, the
potential is
V (z) = Jz =
Iz
A
(17.26)
and thus the potential dierence between the voltage probes is
V =
Ib
A
(17.27)
In the one-dimensional case, the key result is that the potential dierence is independent of the
spacing of the current probes a. From this the resistivity is also independent of a, and is determined
by
=
V A
Ib
(17.28)
As expected, in one dimension, we simply recover the usual expression for a uniform current density.
17.5.5 The van der Pauw Technique
Though the resistivity determinations above are perfectly sensible, they are limited by their rather
strict measurement geometry. In 1958, L.J. van der Pauw proposed a technique for measuring the
ment on a narrow wire sample of cross-sectional
area A. The outer and inner probes are separated
by a distance a, while the inner probes are sepa-
rated by b, {a, b}
A. Current is introduced by
the outer probes, and potential dierence measured
on the inner probes.
resistivity of thin samples of arbitrary shapes. Due to its convenience, it is commonly used tech-
nique to measure the sheet resistance of a material. It can also be used to measure the Hall eect,
which means sheet resistance, carrier type (electron or hole), carrier density, and carrier mobility
can all be determined from a single set of resistance measurements (with the addition of a magnetic
eld for the Hall eect).
Though the full derivation of this technique is somewhat beyond the scope of this lab manual, van
der Pauws original papers on the subject are quite readable, and easily found online.
v
As originally
devised by van der Pauw, one uses an arbitrarily-shaped (but simply connected, i.e., without holes),
thin plate sample of thickness d containing four very small contacts placed on the periphery of the
plate. A schematic of a quasi-rectangular conguration is shown below in Fig. 17.8.
1
2
3
4
Figure 17.8: Contact conguration for a van der Pauw measurement.
The objective of the measurement is to determine the sheet resistance R
s
=/d. Using conformal
mapping techniques related to the analysis above, van der Pauw demonstrated that two character-
istic resistances R
a
and R
b
are sucient to determine the sheet resistance through the following
equation:
v
See Philips Technical Review, vol. 20, pp. 220-224, (1958) and Philips Research Reports, vol. 13, pp. 1-9, (1958).
These articles are available on the course web site in the templates directory.
e
Ra/Rs
+ e
R
b
/Rs
= 1 (17.29)
which can be numerically solved for R
s
. If the thickness of the specimen d is known, may be
calculated. The two characteristic resistances are obtained from the conguration above in the
following manner:
R
a
=
V
43
I
12
=
voltage between contacts 4 and 3
current applied through contact 1 and out of contact 2
(17.30)
R
b
=
V
14
I
23
=
voltage between contacts 1 and 4
current applied through contact 2 and out of contact 3
(17.31)
Specically, one rst sources a current I
12
from contact 1 to contact 2, and measures the voltage
developed V
43
between contacts 4 and 3. The ratio V
43
/I
12
is the characteristic resistance R
a
.
Subsequently, sourcing a current I
23
from contact 2 to contact 3 and measuring the voltage V
14
between contacts 1 and 4 yields the characteristic resistance R
b
. In the rectangular conguration,
this amounts to measuring a single resistance and then rotating the contacts by 90
, as shown in
Fig. 17.9
Figure 17.9: Contact conguration for a van der Pauw measurement showing the determination of the two characteristic
resistances Ra and R
b
. From http://www.eeel.nist.gov/812/hall.html.
In order to use the van der Pauw method, the sample thickness must be much less than the width
and length of the sample. In order to reduce errors in the calculations, it is preferable that the
sample is symmetrical. There must also be no isolated holes within the sample. Further, the con-
tacts must be on the boundary of the sample (or as close to it as possible). Strictly, the contacts
must be innitely small. Practically, they must be as small as possible; any errors given by their
non-zero size will be of the order D/L, where D is the average diameter of the contact and L is
the distance between the contacts.
In addition to this, any leads from the contacts should be constructed from the same batch of wire
17.6 Objective 201
to minimize thermoelectric eects. For the same reason, all four contacts should be of the same
material.
17.6 Objective
In this experiment, you will learn how to measure the electrical resistivity of thin lms and devices
using four-point probe techniques. Additionally, you will learn how to properly utilize signal av-
eraging to reduce measurement uncertainty. Straightforward extensions of these techniques, which
you may explore in additional labs, will allow the measurement of the carrier mobility and concen-
tration via the Hall eect.
Hypothesis:
Electrical measurement uncertainty (noise) can be reduced by repeatedly sampling the quantity of
interest, though intrinsic noise remains, related to the resistance of the device under test. Four point
probe techniques allow measurements of electrical resistivity which can largely eliminate artifacts
due to measurement geometry and non-ideal components.
Taylor[1], Ch. 3, 4, 5, 8
Review of current and resistance, dc circuit analysis (see Appendices).
17.7 Supplies
1. Keithley 220 current source
2. Hewlett-Packard 3457A multimeter
3. Bamalab electrical supply / measurement box
4. decade resistance box
5. Si wafers
6. Nb thin lm on Si substrate
7. banana cables
8. Spring-loaded sample holder
202 17.7 Supplies
Magnet supply
sample voltmeter
sample
current source
Hall voltmeter
Hall
current source
DAC
Figure 17.10: The electrical transport setup.
Figure 17.11: The spring-loaded resistivity sample holder.
17.8.1 Signal averaging and noise
The circuit for this and the following portion of the laboratory is simple: the uctuating voltage
from a decade resistance box will be amplied and fed into a digital oscilloscope with very high
sampling rate. Though there is no external voltage or current source connected to the resistor, you
will be able to see the uctuating (noise) voltage on the resistor as a function of time. While its
time average V ) will be zero, its rms value V
2
) will not.
vi
1. Connect your decade resistor to the input of the PAR 113 preamplier. Set the resistance to
100 k.
2. Set the coupling on each channel (A,B) to ac, with a gain of 1k, and upper and lower cuto
frequencies of 100 kHz and 1 Hz respectively.
3. Connect the PAR 113 output to channel 1 of the LeCroy LT342L oscilloscope.
4. Set the LeCroy to 0.1 sec/div (with 10 divisions, this gives 1 second per trace), with a vertical
scale of 20 mV per division, or as required to t the trace vertically on screen.
Each trace on the screen lasting 1 second will give 5 10
4
points, for a time resolution of 20 s.
Elementary sampling theory dictates that this will result in measurements in the frequency domain
spanning the range from 1 Hz25 kHz. First, you will verify the statistical nature of the noise.
1. Using the on-screen statistics display, tabulate the following quantities after (for example) 20
sweeps:
mean voltage (should be near zero)
standard deviation
rms (should be equal to standard deviation)
amplitude
peak-peak voltage
2. Repeat the measurement for at least 5 resistances in the range of 5100 k.
You may average over many scans if you like, just keep the number of scans the same for all values
of R.
vii
You should nd that the variance of the voltage measured resistance increases:
V
2
) = V
2
rms
=
2
= 4k
B
TRf (17.32)
In this measurement, your expected bandwidth f is set by the upper (100 kHz) and lower (1 Hz)
roll-o lters (so f 157 kHz). Plot
2
versus R, and perform a linear t. Is the slope consistent
vi
Since there is no dc source present, the rms value and standard deviation will be the same, within the osets of
the equipment.
vii
Do not use so many points and samplings that the measurements take an inordinate amount of time, you are
trying to verify the relationship between noise and the external resistance value, not minimize the noise.
with your bandwidth and the room temperature of 295 K? Extract the eective bandwidth, as-
suming a temperature of 295 K for the resistor. Why might your results not agree?
In order to investigate the eects of signal averaging, use the program ScopeExplorer.exe" on the
desktop to capture at least 5 scans of voltage versus time.
1. After opening ScopeExplorer.exe," click Traces on the upper icon tray.
2. Select trace C1, and click Get in ASCII format on the upper icon tray.
3. Choose byte" word size and click OK.
4. Choose time and amplitude" data for export.
5. Do this ve times under nominally identical conditions to accumulate ve separate scans.
6. Plot the average of 2, 3, 4, and 5 scans.
7. Does the standard deviation from these scans decrease as the square root of the number of
scans?
8. If you are familiar with Fourier transformations, perform a transformation on the averaged
data to bring it to the frequency domain. Are there any particular frequencies with larger
rms voltage than others?
viii
9. Since you are looking to verify

N behavior, it is convenient to increase N by a constant
factor for each subsequent measurement.
10. Verify that the standard deviation reduces as the square root of the number of scans.
11. Calculate how many standard deviations away from the mean your peak-peak voltage is after
each set of scans. What is the probability of such a voltage measurement? Compare this to
the number of points measured (5 10
4
per scan).
12. Review the Counting Statistics experiment for further analysis you can perform (e.g., run-
ning average).
17.8.2 Resistivity of thin specimens
Having the knowledge to perform an accurate electrical measurement, we will now consider a prac-
tical problem: measuring the resistivity of a thin specimen. In this case, you will measure the
resistivity of three samples: two doped Si wafers of dierent resistivities, and a thin (10 nm) Nb
lm on Si.
17.8.2.1 Linear Four-Point Probe
In this portion of the experiment, you will use a set of four equally-spaced probes to measure the
resistivity of a Si wafer. For the following experiments, you will use the Keithley 220 current source
and HP3457A multimeter. You no longer need the Bamalab box or decade resistor, you may put
viii
Strictly speaking, you should calculate the auto-correlation function and take the Fourier transformation of that
to get the noise spectral density. For qualitative purposes, just Fourier transforming the raw data will work.
them away. At your work station, you should nd a spring-loaded sample holder (Fig. 17.11) and
samples of Si wafers. Take a piece of the low Si of sucient size to contact four pins in a
straight line, and mount it in the spring-loaded holder:
1. Loosen the four nylon screws (by hand) on the top of the holder until the gold-colored contact
pins no longer touch the bottom plastic plate. You do not need to completely remove the
screws.
2. Using a tweezers, insert the Si wafer into the holder between the lower plate and the
spring-loaded contact pins. Make sure the polished (shiny) side is facing up. Do not touch
the surface of the wafer with your ngers.
3. Press the top plate down until the spring-loaded contact pins are pressing rmly agains the
Si wafer. Do not press too hard! While holding the top plate down, tighten the nylon screws
to hold the plate and contact pins in place.
4. Using a handheld multimeter, verify that the wafer is well-contacted by measuring the resis-
tance between the pins. (Use the wires and banana plugs connected to the pins.) Check that
four pins in a straight line are making contact, the resistance between any two should be at
most several k. If any measurements are in the M range, try tightening the holder slightly
or moving the wafer slightly.
For the linear four-point probe measurement, you wish to source current through the outer of the
four contacts (e.g., blue and green or brown and orange wires), and measure voltage across the
inner two contacts (e.g., grey and black or yellow and red).
1. Connect the Keithley 220 (K220) current source to the wires attached to the outer pair of
the four contacts on your sample. Connect the HP 3457A multimeter to the inner two pins.
2. The maximum suggested current to avoid resistive heating of the sample is such that I
2
R<
0.1 W. Calculate this current using the largest measured resistance.
3. Using the front panel of the K220 current source, enter a current below this maximum value
and press the Enter key. Apply it to the sample pressing the Output button.
4. If the V-limit light is ashing, you will need to increase the voltage limit of the current
source. Do this by pressing the V-limit button, typing a larger value (e.g., 10), and pressing
enter. Repeat until the light no longer ashes. If this does not happen before reaching the
sources limit of 29 V, you most likely have a bad contact.
5. Place the HP 3457A voltmeter in DC V" mode using the front panel buttons.
6. The HP 3457A voltmeter should now register a reading. Increase or decrease the range as
appropriate.
7. Adjust the sample current until your measured voltage is at least 0.1 mV if possible while
staying below your maximum current.
At this point, the ratio of the measured voltage to supplied current gives you the resistivity as
outlined in the discussion above. However, the reading is not yet optimal, and more accuracy could
be achieved by sampling the voltage many times and averaging. The HP 3457A multimeter has
this capability built in. On the front panel, press the NRDGS key, enter the desired number
of readings, and press enter. This will cause the multimeter to make repeated measurements and
display the average result on the front panel. Choose as many averages as necessary to reduce the
uctuation in the voltage to a level you deem acceptable; however, note the decreased response time
of the instrument. You may consider 3 digit accuracy to be sucient. When you are nished:
measure the thickness of the wafer using a calipers, you will need this value to determine resistivity!
V (I) sweep: Another way to increase the accuracy of the measurement is to perform a V (I) sweep.
If the device in question is purely resistive, V (I) will be a straight line, and its slope yields the
resistance. Review quickly the procedure in Sec. 12 for instructions on how to open the appropriate
software and perform a V (I) sweep. WIth this software, you may vary the number of samplings
per point to reduce the noise to levels you deem acceptable.
Repeat the measurement for the high Si sample. Your resistances will be much higher, the
corresponding currents much lower, and a larger number of readings may be required.
17.8.2.2 van der Pauw Technique
The van der Pauw technique, described above, is a powerful technique that allows a measurement
of the resistivity of a thin specimen independent of the placement of the contacts. Moreover, by
performing a set of redundant measurements, one can eliminate the eect of source and meter
osets.
For the van der Pauw measurement, you will need to make four contacts to your sample, preferably
in a square or rectangular conguration.
ix
Choose and mount a piece of the low Si that is of
sucient size to contact four pins in a square or rectangular pattern. Label the contacts 14 in
a counter-clockwise fashion, as shown in Fig. 17.8. We will dene currents and voltages for this
measurement as follows:
I
12
= positive dc current I injected into contact 1 and taken out of contact 2 (17.33)
V
34
= dc voltage measured between contacts 1 and 2 (V
1
V
2
) (17.34)
In principle, only two measurements are required for the van der Pauw technique. However, opti-
mum accuracy and consistency, we will perform a set of 8 measurements, using the following basic
procedure:
1. Source a dc current I such that the power dissipation does not exceed 5 mW. Using the
resistance R between any two opposing leads (e.g., 1-3 or 2-4), this current can be chosen by
ix
The rectangular conguration is not strictly necessary, it just makes things easier to keep track of.
I <1/
200R.
2. Apply the current I
21
and measure the voltage V
34
.
3. Reverse the polarity of the current (I
12
) and measure V
34
.
4. Repeat for the remaining six values (V
41
, V
14
, V
12
, V
21
, V
23
, V
32
).
5. Eight measurements of the voltage yield the following eight values of resistance, all of which
must be positive:
R
21,34
=
V
34
I
21
R
12,43
=
V
43
I
12
(17.35)
R
32,41
=
V
41
I
32
R
23,14
=
V
14
I
23
(17.36)
R
43,12
=
V
12
I
43
R
34,21
=
V
21
I
34
(17.37)
R
14,23
=
V
23
I
14
R
41,32
=
V
32
I
41
(17.38)
Because the second half of this sequence of measurements is redundant, it permits important
consistency checks on measurement repeatability. Consistency following current reversal requires
that:
R
21,34
= R
12,43
R
43,12
= R
34,12
(17.39)
R
32,41
= R
23,14
R
14,23
= R
41,32
(17.40)
The reciprocity theorem requires
R
21,34
+ R
12,43
= R
43,12
+ R
34,21
(17.41)
R
32,41
+ R
23,14
= R
14,23
+ R
41,32
(17.42)
The sheet resistance R
s
=/d can now be determined using the two characteristic resistances
R
a
=
1
4
(R
21,34
+ R
12,43
+ R
43,12
+ R
34,21
) (17.43)
R
b
=
1
4
(R
32,41
+ R
23,14
+ R
14,23
+ R
41,32
) (17.44)
via the van der Pauw equation. A sample worksheet for performing these measurements is provided
on the next page.
x
x
Work of the US government excluded from copyright. From http://www.eeel.nist.gov/812/work.htm
17.9 Data analysis and Discussion 209
17.9 Data analysis and Discussion
17.9.1 Signal averaging and Noise
Perform analyses similar to those for the counting statistics experiment (you may even use the
same template). How can you demonstrate that the noise is random? What is its distribution?
What other techniques might you use to reduce electrical noise (either physical changes or signal
processing) in place of point averaging? What advantages or disadvantages are there?
17.9.2 Resistivity
Determine the resistivity for all three types of measurements, with appropriate uncertainties.
Compare your resistivity values to the stated values. Are you within the specied range? What
value of n is implied for your samples, assuming a single carrier type?
For the linear four-point probe technique, compare your uncertainties when determining R at a
single current to determining R via a linear t to a V (I) characteristic (you should now know how
to determine the uncertainty in t parameters). What advantages does the latter method have?
For the van der Pauw technique, you will need to determine the resistivity from a numerical calcu-
lation. Describe your algorithm and its uncertainty. Discuss also the error implicit in the van der
Pauw measurement. This is discussed in van der Pauws original papers, Philips Technical Review,
vol. 20, pp. 220-224, (1958) and Philips Research Reports, vol. 13, pp. 1-9, (1958).
Discuss the reasons for the eight measurements in the van der Pauw technique.
For a proposal, discuss how the van der Pauw technique can be used to determine carrier mobility
via the Hall eect.
course of all of the one-week experiments you must do dierent numbers of each type of report.
Solving the van der Pauw equations numerically
NIST has published a relatively simple method for accurately solving the van der Pauw equation
for sheet resistance R
s
= /d, given the two appropriate van der Pauw resistance measurements.
Below, we briey reproduce their algorithm.
xi
The sheet resistance R
s
can be obtained from the two measured characteristic resistances
R
A
and R
B
by numerically solving the van der Pauw equation:
e
R
A
/Rs
+ e
R
B
/Rs
= 1 (17.45)
using the following iterative routine:
Set the error limit =0.0005, corresponding to 0.05%.
Calculate the initial value of z
i
, or
z
o
=
2 ln 2
(R
A
+ R
B
)
Calculate the i
th
iteration of
y
i
=
1
exp (z
i1
R
A
)
+
1
exp (z
i1
R
B
)
Calculate the i
th
iteration of z
i
, where
z
i
= z
i1
(1 y
i
) /
R
A
/ exp (z
i1
R
A
) + R
B
/ exp (z
i1
R
A
)
When (z
i
z
i1
)/z
i
is less than , stop and calculate the sheet resistance R
s
=1/z
i
The resistivity is given by = R
s
d, where d is the thickness of the conducting
layer.
What follows is a very basic C program that can be used to solve for the sheet resistance, given
the two appropriate resistances from a van der Pauw measurement. This le should be available
for download on the course web site.
// Fol l ows t he NIST al g or i g t hm from h t t p : //www. e e e l . n i s t . gov /812/samp . htm
#include <s t dl i b . h>
#include <s t di o . h>
#include <math . h>
#define PI 3. 14159
f l oat y_i ( f l oat z_prev , f l oat Ra, f l oat Rb) ;
xi
The box below is reproduced from http://www.eeel.nist.gov/812/samp.htm
f l oat z_i ( f l oat z_prev , f l oat y , f l oat Ra, f l oat Rb) ;
f l oat z_o( f l oat Ra, f l oat Rb) ;
f l oat van_der_pauw( f l oat Ra, f l oat Rb) ;
i nt main ( i nt argc , const char argv [ ] ) {
f l oat Ra, Rb, Rs ;
i f ( argc <=2) {
f p r i n t f ( stdout , " Usage : van_der_pauwRaRb \nReturns : Rs=rho/d\nAl l uni t s : Ohms\n" ) ;
return( 1);
}
Ra = at of ( argv [ 1 ] ) ;
Rb = at of ( argv [ 2 ] ) ;
f p r i n t f ( stdout , "Ra=%g\t Rb=%g\n\n" , Ra, Rb) ;
Rs = van_der_pauw(Ra, Rb) ;
f p r i n t f ( stdout , " Sheet Res i s t ance ( rho/d) Rs=%g\n" , Rs ) ;
return ( 0 ) ;
}
f l oat van_der_pauw( f l oat Ra, f l oat Rb)
{
f l oat Rs , err , y , z , z_prev ;
i nt count =0;
f l oat TOL = 1E8; / how a c c ur a t e l y you want t o s o l v e ! /
z_prev = z_o(Ra, Rb) ;
z=z_prev ;
do {
y = y_i ( z_prev , Ra, Rb) ;
z = z_prev z_i ( z_prev , y , Ra, Rb) ;
e r r = f abs ( ( zz_prev )/ z ) ;
z_prev=z ;
count++;
} while ( e r r >= TOL) ;
Rs = 1. 0 / z ;
return ( Rs ) ;
}
f l oat z_o( f l oat Ra, f l oat Rb) / s t a r t i n g v al ue f o r i t e r a t i o n /
{
f l oat z_o=0;
z_o = 2. 0 l og ( 2 . 0 ) / ( PI (Ra + Rb) ) ;
return ( z_o ) ;
}
f l oat z_i ( f l oat z_prev , f l oat y , f l oat Ra, f l oat Rb) / z_i . . . c a l c u l a t e each i t e r a t i o n /
{
f l oat z_i =0;
z_i = ((1. 0 y)/ PI ) / ( Ra/exp ( PIz_prevRa) + Rb/exp ( PIz_prevRb) ) ;
return ( z_i ) ;
}
f l oat y_i ( f l oat z_prev , f l oat Ra, f l oat Rb)
{
f l oat y_i =0;
y_i = 1. 0 / exp ( PIz_prevRa) + 1. 0 / exp ( PIz_prevRb) ;
return ( y_i ) ;
}
Electrical Resistivity
In order to understand the subtleties of the probe techniques used to measure the resistivity in
this laboratory, it is useful to very quickly review a simple classical model of conduction in metals
and semiconductors. If you are familiar with the concepts of mobility, carrier concentration, and
resistivity, you may proceed to Sec. 17.5.
Electric Current
First, we will consider a homogeneous conductor, whose primary conduction is by electrons, and
later we will generalize our results to the case of semiconductors with both electron and hole con-
duction. If we take a cross section of a conductor, such as a circular wire, an electric current is said
to exist if there is a net ow of charge through this surface. The amount of current is simply the
rate at which charge is owing, the number of charges per unit time that traverse the cross-section.
Strictly speaking, we try to choose the cross-sections for dening charge ow such that the charges
ow perpendicular to that surface. Current is a ux of charge through a wire in the same way that
water ow is a ux of water through a pipe. Qualitatively, this is a reasonable way to think about
electric circuits current always has to ow somewhere, and you dont want an open connection
any more than you would want an open-ended water pipe. Voltage is more like pressure you can
have a voltage even when nothing is owing, it just means there is the potential for ow.
If a net amount of charge Q ows perpendicularly through a particular surface of area A within
a time interval t, we dene the electric current to be simply the rate at which charge passes that
surface:
I
dQ
dt
(17.46)
In other words, current is charge ow per unit time. We should get one thing out of the way right
o the bat: the denition for the current direction is somewhat confusing. The historical denition
is that current ow is dened as the direction that positive charges would be moving.
Of course, at this point we know that usually it is really electrons doing all the moving, but the
denition of electric current has held fast.
Getting Current to Flow
Current in real conductors is due to the (net) motion of microscopic charge carriers. How much
current ows depends on the average speed of these charge carriers, the number of charge carriers
per unit volume (the density of charge carriers), and how much charge is carried by each. But how
do we get charges to ow through a conductor in the rst place?
In order to get a net ow of charges, we need to provide a potential dierence (voltage!
xii
The
presence of a voltage gives rise to an electric eld across the conductor, which in turn causes an
electric force, which accelerates the charges. The eectiveness of a potential dierence to cause a
current depends on the density of charge carriers, their average speed, and microscopic properties
xii
From now on, we will interchangeably use the phrases potential dierence and voltage. From our point of
view, they are the same thing.
of the conductor itself.
The free charges in conductors are extremely numerous and fairly mobile. Inside a normal con-
ductor, like copper, there is a fantastic density of charge carriers, 10
22
electrons per cm
3
![21] So
many, in fact, that they continuously scatter o of each other and the xed atoms in the conductor
(about once every 10
14
sec or so, even in a good conductor!). Typical drift speeds in copper are
10
3
10
4
m/s for moderate electric elds, compared to the speed of random thermal electron
motion of 10
5
m/s.[22] Any particular charge carrier has a hard time getting anywhere. Even
though the charges are mobile, and able to move at fantastic speeds, the time it takes to actually
get anywhere is quite a bit longer than expected. A bit like pachinko.
One result of all these collisions is that the carriers in, e.g., copper, cover huge distances in any
given time interval but have a very small displacement most of their movement is wasted, and
they end up close to where they started out, so their net velocity is very small. Even when we
apply a potential dierence, the net ow of charges is more sluggish than we might expect, due
to all these collisions. The net velocity of charge ow
xiii
we call the drift velocity, v
d
. In normal
conductors, like copper, this drift velocity is more or less proportional to the voltage applied, a
point which we will explore in depth presently.
Drift Velocity and Current
Our conceptual physical picture of current in conductors is basically complete. A voltage induces
an electric eld, which gives the carriers a net velocity in one direction, which is an electric current.
This drift motion along the electric eld is superimposed upon the random thermal motion of the
charge carriers (just like the random thermal motion in an ideal gas). From here, all we need to do
is apply our knowledge of electric forces and elds and kinematics to come up with a relationship
between current, eld, and voltage.
So rst: given a drift velocity v
d
, through a conductor of cross section A, what is the current? The
number of charges that ow through our cross section A in the time dt is just the free charge which
is physically close enough to reach the surface A within that time. Those charges close enough
must cover the distance dx in the time dt. Since the average speed of the carriers is v
d
, then we
must have dx=v
d
dt. This is illustrated schematically in Figure 17.12.
The number of charges which cross the surface A, those close enough to reach it in a time dt, is
just the number contained within the volume A dx, or Av
d
dt. A bit more mathematically, we can
write this:
xiii
Distinct from and not to be confused with the random thermal motion, see below.
q
v
d
t
v
d
x
A
Figure 17.12: A small piece of a conductor of cross-
sectional area A. The charge carriers move with a speed
v
d
, and are displaced by dx = v
d
dt in a time interval
dt. The number of carriers in a section of length dx is,
on average, nAv
d
dt, where n is the density of the charge
carriers.
number of charge carriers N = charge density volume (17.47)
= charge density area distance covered in timedt (17.48)
= nAdx = nAv
d
dt (17.49)
Here we have used n to represent the number of charges per unit volume, the carrier density. The
total amount of charge is the number of charge carriers times how much charge each one carries,
which well call q. The current then is just the total amount of charge, Nq divided by the total
amount of time, dt:
I =
dQ
dt
=
Nq
dt
=
nqAv
d
dt
dt
= nqAv
d
(17.50)
We can see that the drift velocity and resulting current are larger when the carriers carry more
charge q, or when their mass is small. However, it would be nice to have expressions that didnt
directly involve the cross-sectional area of the conductor, so we can calculate general properties
independent of any particular conductor shape or size. For this reason, it is common to introduce
current density, J, which is just the current per unit area. Rewriting Eq. 17.50 in terms of current
density, we come up with a simpler and more general expression:
J
I
A
= nqv
d
(17.51)
Now we can calculate the current density for any given material of arbitrary geometry, and later
specify a cross-sectional area to determine absolute currents.
Resistance and Ohms Law
From Equation 17.50, we saw that the current through a conductor can be expected to scale with
the drift velocity. You might expect that the eect of increasing the applied voltage across a con-
ductor V is to increase the drift velocity. This is basically true, but justifying that statement will
require a few more steps.
More accurately, the presence of a potential dierence between two points on the conductor means
that those two points are at dierent potential energies. Recall that negative charges want to move
from regions of lower potential to regions of higher potential. In a conductor, even when a current
ows, the charges like to spread out as evenly as possible. This even and moving distribution of
charge gives rise to a uniform electric eld. If the potential dierence V is applied over some
distance l, and the electric eld is uniform, we know that the electric eld along the length of the
conductor must be given by:
E =
V
l
(17.52)
The presence of the electric eld causes an acceleration of the charge carriers:
a =
F
e
m
=
q
m
E (17.53)
Thus the acceleration of the charge carriers depends only on the electric eld and their charge-mass
ratio, q/m, about 1.7610
11
C/kg for electrons. In order to gure out how much current will ow
for a given potential dierence, we need to nd a way to take into account the dissipative eect of
all the collisions the carriers are constantly undergoing. In a sense, the collection of charge carriers
is a bit like an ideal gas, and our treatment here is reminiscent of an ideal gas law derivation. The
analogy is a close one the innumerable electrons in a conductor are often called an electron gas.
Drift Velocity and Collisions
If we assume the charge carriers are electrons, of mass m
e
(and charge e), then each has an
average momentum p=m
e
v
d
. We expect on average that each collision an electron experiences will
completely destroy all forward momentum they are stopped cold by every single collision. This
makes some sense, since most of the collisions will be with the atoms making up the conductor,
which are very heavy compared to electrons, rather than with other electrons. If all forward mo-
mentum is destroyed, then the electron is left with only its random thermal motion. If there were
no electric force present to accelerate the electrons, the random thermal motion of all the electrons
will cancel out, and there is no net ow or current.
We can easily nd the thermal velocity of the carriers just like we do for an ideal gas the thermal
energy of the electrons is
3
2
k
B
T, where k
B
is Boltzmanns constant, and we equate this to the
carriers kinetic energy:
3
2
k
B
T =
1
2
mv
2
th
(17.54)
=v
th
=
3k
B
T
m
10
5
m/s (at 295 K) (17.55)
Here we use v
th
to specify the thermal velocity distinctly from the electric-eld-induced drift veloc-
ity. As it turns out, the thermal velocity typically greatly exceeds the drift velocity (by ten million
times or so!) the acceleration of the carriers by the electric eld induces only a tiny velocity
compared to that given by the random thermal motion of the carriers. Again, this is what leads to
carriers covering huge distances but having very small displacements. The overall motion is terribly
chaotic, and even fairly large electric elds only alter the carrier velocity in conductors by parts
per million at best. Still, the random thermal velocities do not contribute to the electric current,
xiv
it is only the tiny eld-induced drift velocity that gives rise to electric current.
We should also keep in mind that the collisions the carriers undergo are not continuous, but happen
one after another with some average time between them .
xv
In that time interval, the electron
loses its momentum m
e
v
d
due to a collision, and thereafter regains it due to the action of electric
eld present, only to lose it again about seconds later. As stated above, the presence of the
electric force F
e
gives the electron an acceleration a=F
e
/m
e
, which allows it to regain its former
drift velocity. From kinematics, we would expect a mean displacement v
d
a.
xvi
The starting and stopping motion of the carriers gives us an average rate at which the electrons
are losing momentum due to the collisions and associated impulse forces. We can straightforwardly
nd this momentum change as:
_
p
t
_
loss
=
m
e
v
d
(17.56)
Once the scattering event is over, the electron regains momentum through the action of the electric
force caused by the electric eld. We can easily write down the momentum gained up until the
next collision:
_
p
t
_
gain
= F
e
= qE = eE (17.57)
Now, the total momentum loss has to equal the total momentum gain for there to be a steady
state. If this were not true, the momentum would quickly build up, and the whole wire would start
to move! So we must impose conservation of momentum:
_
p
t
_
loss
=
_
p
t
_
gain
(17.58)
m
e
v
d
= eE
x
(17.59)
v
d
=
e
m
e
E (17.60)
xiv
They do give rise to electrical noise, however.
xv
For Cu, we can estimate[21] 2 10
14
s.
xvi
Depending on the method of derivation, there may be a factor of 2 in this expression, but the physics is the same.
Really, this is just an application of Newtons laws p/t is a force, and the equations above are
also essentially a force balance between the electric force and the impulse force due to the collision.
Now we have an expression for the average drift velocity of electrons owing along the wire, in
terms of the average time between carrier collisions:
v
d
=
e
m
e
E (17.61)
The minus sign makes sense here, by the way. Since electrons are negatively charged, they move in
the opposite direction that the electric eld lines point. It is also reassuring that the drift velocity
increases as increases, since more time between collisions means more time spent accelerating,
and that in principle lighter carriers would have a higher velocity since they are more easily accel-
erated. Finally, the proportionality with the electric eld is what we expect.
For typical metals, we can estimate[22] drift velocities of about 510
3
m/s for a moderate electric
eld of 1 V/m, about eight orders of magnitude below the thermal velocity! Really, the eect of the
electric eld is quite negligible in one sense, though it has profound consequences.
Mean Free Path and Mobility
Instead of dealing with the mean time between collisions, we could just as easily have started with
the mean distance that electrons travel before undergoing a collision.
xvii
This quantity is known as
the mean free path,
mfp
, and it has essentially the same meaning as it does in the kinetic theory
of gasses. The shorter the time between collisions, the smaller the mean free path, and vice versa.
The mean time and mean free path are easily related through kinematics:
mfp
= (v
d
+ v
th
) v
th
(17.62)
Here we are considering the total distance covered not just the net displacement, so we need to use
the total velocity, v
d
+v
th
. For the last relationship, we have made use of the fact that v
th
v
d
.
What this means is that the mean distance (and mean time) between collisions does not really
depend on the applied electric eld, but really only comes from the random thermal motion of the
carriers.
The proportionality constant between drift velocity and electric eld in Eq. 17.61 is commonly
called the carrier mobility, which is just what it sounds like. In this case, we write v
d
=E, where
is the mobility:
v
d
= E with =
q
m
(17.63)
xvii
Here we do mean the distance covered between collisions, not the displacement
From the units of (m
2
/Vs) and E (N/C or V/m), we can see that mobility is a quantity that
tells us how far a charge is able to move per second per unit of electric eld (V/m). Now we have
a nice expression for exactly what we mean by mobility, rather than just a vague notion.
Current, Electric Field, and Voltage
Plugging Eq. 17.61 into Eq. 17.51, we nd the relationship between current density and electric
eld, Ohms law:
J =
I
A
= nqv
d
= ne
eE
m
e
=
ne
2
m
e
E
1
E (17.64)
In the end, it turns out that current density (or current) and electric eld are simply proportional.
We could almost have guessed this in the rst place, but now we have a formal relationship between
the two, and we even know the constant of proportionality. Typically, we dene a new quantity
, the electrical resistivity, which is the constant of proportionality between current density and
electric eld:
xviii
=
m
e
ne
2
=
1
ne
=
1
(17.65)
The conductivity is simply the inverse of the resistivity. Resistivity represents the eectiveness
with which a given electric eld or potential dierence causes a current to ow, and is a (strongly)
material-dependent property it is a measure of the resistance of a material to current ow. We
see that the resistivity gets larger when the time between electron collisions gets smaller, just
as we would expect, and it gets larger when we increase the density of free carriers. Similarly,
resistivity an mobility are inversely proportional. We can go further in our analysis by noting that
the potential dierence and electric eld are simply related in a conductor by E = V/l, which
leads us to:
J =
I
A
=
1
V
l
or V =
l
A
I = lJ (17.66)
In other words, we nd J I V the current ow in a conductor is proportional to the
magnitude of the applied voltage, and the amount of current one gets for a particular applied voltage
depends on the conductors resistivity and geometry. We can make this simpler by introducing a
new constant of proportionality R=
l
A
. This, along with the denition of current density (J =I/A),
will allow us to relate I and V directly. This new constant of proportionality R between
I and V is known as the resistance of the conductor, and it allows us to connect V and
I in the traditional form known as Ohms
xix
law:
xviii
We will use a slightly dierent rho character for resistivity, , to distinguish it from the one we use for mass
density, .
xix
After Georg Simon Ohm (17891854) a German physicist who rst found the relationship between current,
voltage, and resistance.
V = IR or I =
V
R
or R =
V
I
(17.67)
Transport in Semiconductors
The primary dierence between metals and semiconductors, from our point of view, is that semi-
conductors can have electrical conduction by both electrons and holes.
xx
In practical semiconduct-
ing devices, the carrier concentrations are such that the approximation of a small drift velocity
compared the the thermal velocity holds, and our model of collision-dominated conduction above
well-describes both electron and hole motion.
We can consider the electron and hole contributions to the conductivity separately. If an electron
meets a hole in the semiconductor, the electron will simply annihilate the hole (i.e., occupy the
empty state that the hole represents), and neither will contribute to conduction. For this reason,
we may consider the two channels of conductivity to be essentially non-interacting (other than the
presence of one inuencing the mean scattering time of the other), and treat them as two parallel
conductors within the same material. We can then add the electron and hole resistivities in the
same fashion that we add resistors in parallel, since the geometrical factors relating resistivity to
resistance are the same for both. That is, we add the resistivities inversely (or simply add the
conductivities). If we denote the electron density and mobility as n and
n
, and the hole density
and mobility as p and
p
, this leads to a total resistivity
tot
1
tot
=
1
n
+
1
p
= ne
n
+ pe
p
(17.68)
tot
=
1
ne
n
+ pe
p
(17.69)
If one species of carrier dominates the conductivity either by sheer numbers or by a vastly larger
mobility the expression reduces to that of a normal conductor above, which is the situation we
will encounter. Which type of carrier is dominant cannot be determined by a simple resistivity
measurement, being insensitive to the sign of the charge carrier. The Hall eect, discussed below,
can make this determination. If we simply measure the resistivity of a semiconductor without
regard to whether both charge carriers play a signicant role or not, we can extract an eective
mobility
e
and carrier concentration N:
tot

1
Ne
e
(17.70)
N
e
n
n
+ p
p
(17.71)
xx
A hole is the conceptual opposite of an electron, and describes the lack of an electron at a position where one
could exist. It is not the same as a positron.
For a doped semiconductor, one can show that the product np is constant, np = n
2
i
where n
i
is
the intrinsic carrier concentration. Without doping, the carrier concentrations must be equal, and
n=p=n
i
. Except for very high carrier densities, approaching that of a metal, n and p are highly
temperature-dependent, increasing as temperature increases. This is due to the fact that in a
lightly-doped semiconductor the concentration of free carriers of either type is strongly determined
by thermal activation. The mobility shows a strong temperature dependence as well, with mobility
decreasing strongly as temperature increases. For pure silicon, n
i
1.510
10
cm
3
at 300 K, about
12 orders of magnitude below that of copper. Doped silicon can have n
i
10
13
10
18
cm
3
, above
n
i
10
18
cm
3
, one usually considers the semiconductor so highly doped that it is for all intents
and purposes a metal.
Measuring resistive devices
Sourcing Voltage
A current can only be maintained in a closed circuit by a source of electrical energy. The simplest
way to generate a current in a circuit is to use a voltage source, such as a battery. A voltage source
essentially raises or lowers the potential energy of charges that pass through it. The amount of
energy gained per charge that passes through a device is the potential dierence that the voltage
supplies, V , measured in Joules per Coulomb (J/C), i.e., Volts (V). Though voltage is strictly an
energy per unit charge, it is often useful to think of a voltage as a pressure of sorts, which tries
to force charges through an electric circuit. Just like hydrostatic pressure, the presence of a voltage
does not necessarily lead to a current, this only occurs when a completed circuit is present. In this
way of thinking, a voltage source is a sort of generalized power supply which can be thought of as a
charge pump that tries to force charges to move within an electric eld inside the source. Many
batteries, for instance, are electron pumps in which negatively charged electrons move opposite
to the direction of the electric eld. In an idealized voltage source, the output terminals provide a
constant potential dierence V , and can pump any amount of charge through any closed circuit
connected to the output terminals.
Figure 17.13: A real voltage source provides a voltage
V , but has an internal resistance r. The actual output
voltage developed at its terminals depends on r and the
resistance of the external circuit connected to the battery.
Real voltage sources, however, always have internal resistances, resulting in parasitic voltage losses
within the source itself, and they have power limits which restrict the amount of current that can
be sourced. In general, we can model a real voltage source as an ideal voltage source V in series
with an internal resistance r, as illustrated in Figure 17.13. The eect of the internal resistance
is clear: as soon as an external load is connected to the voltage source and a current ows, the
voltage at the battery terminals is always less than that of the ideal internal source. The only way
to realize the ideal voltage of a source is if it drives no current hardly useful for our purposes.
As a concrete example, consider the circuit in Figure 17.14, a voltage source V with internal
resistance r connected to an external resistor R.
xxi
If we neglect the internal resistance of the
battery, the potential dierence across the battery terminals is V .
Figure 17.14: A voltage source V with internal resis-
tance r connected to an external resistor (load) R.
Once the external (load) resistance R is connected to the source, a single current I is produced in
this single-loop circuit. Conservation of energy (a.k.a. Kirchhos voltage law) requires the sum of
potential dierences around the entire circuit be zero:
0 = V Ir IR (17.72)
Thus, the voltage delivered to the external load resistance R is only
V
load
= V Ir = IR = V
R
r + R
(17.73)
This makes it clear that the voltage across the load is the same as the ideal voltage V only when
the current is zero. This is why another name for the rated voltage is the open-circuit voltage
rated and actual voltages are only the same for a real voltage source when nothing is connected
and no current ows. For completeness, given a load resistance R, internal resistance r, and an
ideal open-circuit source voltage V , we can also determine the current:
I =
V
R + r
(17.74)
Clearly, the current delivered by the battery through the resistor depends on both the resistors
value and the internal resistance of the battery. If R r, of course we need not worry about the
xxi
We are assuming, for now, that wires connecting to the source have no resistance.
internal resistance of the battery, and this is the regime we prefer to operate in. In a nutshell:
voltages sources like high load resistances, compared to their internal resistance.
xxii
Sourcing Current
A current source is nothing more than a device that delivers and absorbs a constant current,
sourcing and sinking a constant number of charges per unit time. An ideal current source (which
exists only on paper) delivers a constant current to any closed circuit connected to its output
terminals, no matter what the voltage or load resistance. Though a battery provides a simple
example of a voltage source, there is no correspondingly simple realization of a current source.
Circuit diagram symbol for a current source:
We can approximate a current source, however, with a single battery and resistor. In the circuit
of Fig. 17.15, a battery with internal resistance connected to a load resistor, the current through
the load is given by Eq. 17.74. If we make the load resistor very small (or equivalently, make the
internal resistance of the battery very large), r R
load
, then the current through the load resistor
is I V/r. This does provide a roughly constant current, but the power loss in the internal
resistor will be severe, and it is generally impractical to construct a current source in this way (that
is not to say that it is not very commonly done anyway, however).
How more realistic constant current sources work internally is a bit beyond the scope of our discus-
sion. However, that does not prevent us from seeing how they behave when connected to a circuit.
In the same way that a real voltage source can be considered an ideal voltage source in series with
a resistor, a real current source can be considered an ideal current source in parallel with a resistor,
as shown in Fig. 17.15.
Figure 17.15: A real current source can be considered
as an ideal current source in parallel with an internal
resistance r. The internal resistance steals some of the
current, depending on the value of the load resistor.
If the internal resistance is very large, almost all of the current goes through the load, and the
current source is nearly ideal. If the load resistance becomes comparable to the internal resistance,
however, a signicant portion of the current takes the parasitic path (I
p
in the gure) through
the internal resistance, and the source is no longer close to ideal. The current through the load
xxii
Good laboratory voltage sources can have internal resistances well below 1 .
resistance is easily calculated using charge and energy conservation (a.k.a., Kirchhos current and
voltage laws):
I
load
= I
r
r + R
(17.75)
The current through the load is independent of the load resistance R and nearly equal to the source
current I when r R
load
. In other words, current sources want low load resistances, in contrast
to voltage sources. This brings up one answer to a common question: is it better to source current
or voltage? If the load you are trying to source has a large resistance, as might be the case for a
tunneling device, sourcing voltage is generally better. If the load is small, as is usually the case for
all-metal GMR devices, sourcing current is generally better.
xxiii
Measuring Voltage
A voltmeter is just what it sounds like a device that measures voltage, or potential dierence,
between two points. A typical voltmeter has two input terminals, and one connects wires from
these input terminals to the points within a circuit between which one wants to know the potential
dierence. If we wish to measure the potential dierence across a particular component in a circuit,
we connect the voltmeter in parallel with that component.
Circuit diagram symbol for a voltmeter:
V
Of course, the idea is to measure the potential dierence while disturbing the circuit as little as
possible. For this reason, voltmeters have very high internal resistances, such that their current
draw is negligible, as shown in Fig. 17.16. An ideal voltmeter probes the potential dierence
between its inputs, but since no current ows through it, it does not aect the circuit. Thus, an
ideal voltmeter should be connected in parallel with the device to be measured.
Real voltmeters have a nite internal resistance, and their current draw is not zero What the
voltmeter really measures then is not just the load, but the equivalent resistance of the load in
parallel with its own internal resistance r. Put another way, the voltmeter forms a current divider
with the load, and shunts part of the current through the load. The voltmeter shunting part of
the current obviously leads to inaccurate results, and the measured voltage drop across the resistor
is no longer IR
load
like we expect. If we assume there is a current I in the wire leading to the
resistor, we can readily calculate the voltage measured by the voltmeter:
V
measured
= IR
eq
=
rR
load
r + R
load
I =
IR
load
1 +
R
load
r
=
V
expected
1 +
R
load
r
(17.76)
The ratio between the measured voltage and the expected value is 1/(1 + R
load
/r), which tells us
two things. First, the measured value is always smaller than the true value, since 1/(1 +R
load
)1.
xxiii
For sources, internal resistance is often called output resistance. Good laboratory current sources can have
internal resistances above 10
14
, while good laboratory voltage sources can have internal resistances well below 1 ,
so with good equipment either I or V can usually be sourced for most common measurements without issues. Noise
is often what actually determines which is used, but even so, the rule of thumb is still useful.
Figure 17.16: (a) An ideal voltmeter has an innite internal resistance, and no current ows through it. Hence, it measures
the true voltage drop across the resistor, V =IR. (b) A real voltmeter has a nite internal resistance r, and forms a voltage
divider with the load resistor. Some current ows through the voltmeter itself if R
load
is comparable to r, and the measured
voltage is less than the true voltage on the resistor.
Second, so long as the load resistor is small compared to the internal resistance of the meter,
R
load
r, the measured and expected values will be very close. Given the enormous internal
resistance of most modern voltmeters, this is usually the case, but one must still exercise caution.
Using a meter with insucient internal resistance is known as measuring the meter, and is
something you will encounter in your laboratory experiments.
xxiv
Measuring Current
An ammeter is the device that measures current, and it behaves rather dierently than a voltmeter.
Measuring the ow of charge has similarities with measuring the ow of uids. A ow meter
measures uid ow by allowing the uid of interest to pass through it. Similarly, an ammeter
measures charge ow by allowing current to pass through it. Ammeters therefore connect in series
with the device to be measured, but one should be aware that real ammeters have an internal
resistance which is thus introduced in series with the load. Ammeters typically have tiny internal
resistances compared to the devices of interest, and current ows readily through them. If an
ammeter is connected incorrectly in parallel with the load, it will create current divider (parallel
resistor network) with the load resistor. The small internal resistance of the ammeter can shunt
most of the current from the load resistor, and an improper measurement results. Since the ammeter
resistance is small, it can be connected in series with the load, and the voltage drop across the
ammeter is usually negligible it measures the current without disturbing the circuit.
Circuit diagram symbol for an ammeter:
A
xxiv
Good laboratory voltmeters can have internal resistances on the order 10
10
or more. For meters, internal
resistance is often called input resistance.
Figure 17.17: A real ammeter measures
current passing through it, but introduces a
series resistance r, creating an additional
voltage burden on the source.
A simple ammeter can be constructed using a precise resistor and a good voltmeter, as shown
in Fig. 17.18. A precise resistor placed in series with the device to be measured (in place of the
ammeter in Fig. 17.17, for instance), and a voltmeter measures the voltage drop across this precise
resistor.
Figure 17.18: A simple ammeter can be constructed
from a precise resistor and a good voltmeter. Since the
value of the resistance is known, the measured voltage
drop across it yields the current.
Since the value of the resistor is known precisely, the measured voltage drop across it yields the
current via Ohms law:
I =
V
measured
R
precise
(17.77)
In this way currents can be measured reasonably accurately, but this is far from an ideal ammeter.
First, this technique of current measurement brings in all the non-idealities associated with real
voltmeters as discussed above. Second, placing a resistor within the circuit of interest introduces
an additional voltage drop, which can aect other components. Care must be exercised when using
this technique. The precise resistor can be chosen carefully as not to introduce a suciently large
voltage drop to alter the circuit too much, the voltages on other components in the circuit must
be independently measured to take this eect into account, or the circuit must be designed from
scratch to account for this additional voltage drop. More accurate instruments, such as current
preampliers, allow far more precise measurements with very low internal resistances. While their
internal complexity is beyond the scope of the current discussion, they are in principle used just as
as the ammeters discussed above.
xxv
xxv
Good laboratory current preampliers can have internal resistances far less than 1 , depending on the level of
Finally, we leave you with a few more rules of thumb: voltmeters have a high internal resistance
and connect in parallel with the device to be measured; ammeters have a low internal resistance
and connect in series with the device to be measured.
current being measured.
17
Reference Data
17.11 Periodic Table of the Elements
On the following page, a freely-distributable periodic table from http://www.nist.gov/physlab/
data/periodic.cfm [23].
227
S
o
l
i
d
s
A
r
t
i
f
i
c
i
a
l
l
y
P
r
e
p
a
r
e
d
L
i
q
u
i
d
s
G
a
s
e
s

5
8
C
e
C
e
r
i
u
m
1
4
0
.
1
1
6
5
.
5
3
8
7
A
t
o
m
i
c
N
u
m
b
e
r
S
y
m
b
o
l
N
a
m
e
G
r
o
u
n
d
-
s
t
a
t
e
C
o
n
f
i
g
u
r
a
t
i
o
n
G
r
o
u
n
d
-
s
t
a
t
e
L
e
v
e
l
I
o
n
i
z
a
t
i
o
n
E
n
e
r
g
y

(
e
V
)
B
a
s
e
d

u
p
o
n

1
2
C
.

(
)

i
n
d
i
c
a
t
e
s

t
h
e

m
a
s
s

n
u
m
b
e
r

o
f

t
h
e

m
o
s
t

s
t
a
b
l
e

i
s
o
t
o
p
e
.

A
t
o
m
i
c
W
e
i
g
h
t
P

E

R

I

O

D

I

C

T

A

B

L

E
A
t
o
m
i
c

P
r
o
p
e
r
t
i
e
s

o
f

t
h
e

E
l
e
m
e
n
t
s

2
9 C
u
C
o
p
p
e
r
6
3
.
5
4
6
7
.
7
2
6
4

1
1
N
a
S
o
d
i
u
m
2
2
.
9
8
9
7
7
0
5
.
1
3
9
1

1
2 M
g
M
a
g
n
e
s
i
u
m
2
4
.
3
0
5
0
7
.
6
4
6
2

1
3
A
l
A
l
u
m
i
n
u
m
2
6
.
9
8
1
5
3
8
5
.
9
8
5
8

1
4
S
i
S
i
l
i
c
o
n
2
8
.
0
8
5
5
8
.
1
5
1
7

1
5
P
P
h
o
s
p
h
o
r
u
s
3
0
.
9
7
3
7
6
1
1
0
.
4
8
6
7

1
6
S
S
u
l
f
u
r
3
2
.
0
6
5
1
0
.
3
6
0
0

1
7
C
l
C
h
l
o
r
i
n
e
3
5
.
4
5
3
1
2
.
9
6
7
6

1
8
A
r
A
r
g
o
n
3
9
.
9
4
8
1
5
.
7
5
9
6

1
2
S
1
/
2
H
H
y
d
r
o
g
e
n
1
.
0
0
7
9
4
1
3
.
5
9
8
4

4
B
e
B
e
r
y
l
l
i
u
m
9
.
0
1
2
1
8
2
9
.
3
2
2
7

3
7 R
b
R
u
b
i
d
i
u
m
8
5
.
4
6
7
8
4
.
1
7
7
1

5
5 C
s
C
e
s
i
u
m
1
3
2
.
9
0
5
4
5
3
.
8
9
3
9

4
2 M
o
M
o
l
y
b
d
e
n
u
m
9
5
.
9
4
7
.
0
9
2
4

4
1 N
b
N
i
o
b
i
u
m
9
2
.
9
0
6
3
8
6
.
7
5
8
9

8
6 R
n
R
a
d
o
n
(
2
2
2
)
1
0
.
7
4
8
5

7
4
W
T
u
n
g
s
t
e
n
1
8
3
.
8
4
7
.
8
6
4
0

4
3
T
c
T
e
c
h
n
e
t
i
u
m
(
9
8
)
7
.
2
8

7
5 R
e
R
h
e
n
i
u
m
1
8
6
.
2
0
7
7
.
8
3
3
5

4
4 R
u
R
u
t
h
e
n
i
u
m
1
0
1
.
0
7
7
.
3
6
0
5

7
6 O
s
O
s
m
i
u
m
1
9
0
.
2
3
8
.
4
3
8
2

4
5 R
h
R
h
o
d
i
u
m
1
0
2
.
9
0
5
5
0
7
.
4
5
8
9

7
7
I
r
I
r
i
d
i
u
m
1
9
2
.
2
1
7
8
.
9
6
7
0

4
6 P
d
P
a
l
l
a
d
i
u
m
1
0
6
.
4
2
8
.
3
3
6
9

7
8
P
t
P
l
a
t
i
n
u
m
1
9
5
.
0
7
8
8
.
9
5
8
8

4
7 A
g
S
i
l
v
e
r
1
0
7
.
8
6
8
2
7
.
5
7
6
2

7
9 A
u
G
o
l
d
1
9
6
.
9
6
6
5
5
9
.
2
2
5
5

4
8 C
d
C
a
d
m
i
u
m
1
1
2
.
4
1
1
8
.
9
9
3
8

8
0 H
g
M
e
r
c
u
r
y
2
0
0
.
5
9
1
0
.
4
3
7
5

6
0 N
d
N
e
o
d
y
m
i
u
m
1
4
4
.
2
4
5
.
5
2
5
0

6
2
S
m
S
a
m
a
r
i
u
m
1
5
0
.
3
6
5
.
6
4
3
7

6
3 E
u
E
u
r
o
p
i
u
m
1
5
1
.
9
6
4
5
.
6
7
0
4

6
4 G
d
G
a
d
o
l
i
n
i
u
m
1
5
7
.
2
5
6
.
1
4
9
8

6
5
T
b
T
e
r
b
i
u
m
1
5
8
.
9
2
5
3
4
5
.
8
6
3
8

6
1
P
m
P
r
o
m
e
t
h
i
u
m
(
1
4
5
)
5
.
5
8
2

6
6 D
y
D
y
s
p
r
o
s
i
u
m
1
6
2
.
5
0
0
5
.
9
3
8
9

6
7 H
o
H
o
l
m
i
u
m
1
6
4
.
9
3
0
3
2
6
.
0
2
1
5

6
8
E
r
E
r
b
i
u
m
1
6
7
.
2
5
9
6
.
1
0
7
7

6
9 T
m
T
h
u
l
i
u
m
1
6
8
.
9
3
4
2
1
6
.
1
8
4
3

4
9
I
n
I
n
d
i
u
m
1
1
4
.
8
1
8
5
.
7
8
6
4

5
0
S
n
T
i
n
1
1
8
.
7
1
0
7
.
3
4
3
9

5
1
S
b
A
n
t
i
m
o
n
y
1
2
1
.
7
6
0
8
.
6
0
8
4

5
2
T
e
T
e
l
l
u
r
i
u
m
1
2
7
.
6
0
9
.
0
0
9
6

5
3
I
I
o
d
i
n
e
1
2
6
.
9
0
4
4
7
1
0
.
4
5
1
3

8
1
T
l
T
h
a
l
l
i
u
m
2
0
4
.
3
8
3
3
6
.
1
0
8
2

8
2
P
b
L
e
a
d
2
0
7
.
2
7
.
4
1
6
7

8
3
B
i
B
i
s
m
u
t
h
2
0
8
.
9
8
0
3
8
7
.
2
8
5
5

8
4 P
o
P
o
l
o
n
i
u
m
(
2
0
9
)
8
.
4
1
4

8
5
A
t
A
s
t
a
t
i
n
e
(
2
1
0
)

5
8
C
e
C
e
r
i
u
m
1
4
0
.
1
1
6
5
.
5
3
8
7

5
9
P
r
P
r
a
s
e
o
d
y
m
i
u
m
1
4
0
.
9
0
7
6
5
5
.
4
7
3

7
0
Y
b
Y
t
t
e
r
b
i
u
m
1
7
3
.
0
4
6
.
2
5
4
2

9
0
T
h
T
h
o
r
i
u
m
2
3
2
.
0
3
8
1
6
.
3
0
6
7

9
2
U
U
r
a
n
i
u
m
2
3
8
.
0
2
8
9
1
6
.
1
9
4
1

9
3 N
p
N
e
p
t
u
n
i
u
m
(
2
3
7
)
6
.
2
6
5
7

9
4
P
u
P
l
u
t
o
n
i
u
m
(
2
4
4
)
6
.
0
2
6
0

9
5
A
m
A
m
e
r
i
c
i
u
m
(
2
4
3
)
5
.
9
7
3
8

9
6
C
m
C
u
r
i
u
m
(
2
4
7
)
5
.
9
9
1
4

9
1
P
a
P
r
o
t
a
c
t
i
n
i
u
m
2
3
1
.
0
3
5
8
8
5
.
8
9

9
7
B
k
B
e
r
k
e
l
i
u
m
(
2
4
7
)
6
.
1
9
7
9

9
8
C
f
C
a
l
i
f
o
r
n
i
u
m
(
2
5
1
)
6
.
2
8
1
7

9
9
E
s
E
i
n
s
t
e
i
n
i
u
m
(
2
5
2
)
6
.
4
2

1
0
0
F
m
F
e
r
m
i
u
m
(
2
5
7
)
6
.
5
0

1
0
1
M
d
M
e
n
d
e
l
e
v
i
u
m
(
2
5
8
)
6
.
5
8

1
0
2
N
o
N
o
b
e
l
i
u
m
(
2
5
9
)
6
.
6
5

1
0
5

1
0
7

1
0
6

1
0
8

1
0
9

1
1
1

1
1
0

1
1
2
D
b
D
u
b
n
i
u
m
(
2
6
2
)
S
g
S
e
a
b
o
r
g
i
u
m
(
2
6
6
)
H
s
H
a
s
s
i
u
m
(
2
7
7
)
B
h
B
o
h
r
i
u
m
(
2
6
4
)
M
t
M
e
i
t
n
e
r
i
u
m
(
2
6
8
)
U
u
n
U
n
u
n
n
i
l
i
u
m
(
2
8
1
)
U
u
u
U
n
u
n
u
n
i
u
m
(
2
7
2
)
1
s

1
1
4

1
1
6

3
1
s
2
2
s
L
i
L
i
t
h
i
u
m
6
.
9
4
1
5
.
3
9
1
7

1
0
N
e
N
e
o
n
2
0
.
1
7
9
7
2
1
.
5
6
4
5

2
H
e
H
e
l
i
u
m
4
.
0
0
2
6
0
2
2
4
.
5
8
7
4

9
O
O
x
y
g
e
n
1
5
.
9
9
9
4
1
3
.
6
1
8
1

8
F
F
l
u
o
r
i
n
e
1
8
.
9
9
8
4
0
3
2
1
7
.
4
2
2
8

7
N
N
i
t
r
o
g
e
n
1
4
.
0
0
6
7
1
4
.
5
3
4
1

6
C
C
a
r
b
o
n
1
2
.
0
1
0
7
1
1
.
2
6
0
3

5
B
B
o
r
o
n
1
0
.
8
1
1
8
.
2
9
8
0

5
7
L
a
L
a
n
t
h
a
n
u
m
1
3
8
.
9
0
5
5
5
.
5
7
6
9

8
9 A
c
A
c
t
i
n
i
u
m
(
2
2
7
)
5
.
1
7

7
1
L
u
L
u
t
e
t
i
u
m
1
7
4
.
9
6
7
5
.
4
2
5
9

1
0
3 L
r
L
a
w
r
e
n
c
i
u
m
(
2
6
2
)
4
.
9

?

8
7
F
r
F
r
a
n
c
i
u
m
(
2
2
3
)
4
.
0
7
2
7

8
8
R
a
R
a
d
i
u
m
(
2
2
6
)
5
.
2
7
8
4

1
0
4
?
R
f
R
u
t
h
e
r
f
o
r
d
i
u
m
(
2
6
1
)
6
.
0

?

7
2
H
f
H
a
f
n
i
u
m
1
7
8
.
4
9
6
.
8
2
5
1

4
0
Z
r
Z
i
r
c
o
n
i
u
m
9
1
.
2
2
4
6
.
6
3
3
9

3
9
Y
Y
t
t
r
i
u
m
8
8
.
9
0
5
8
5
6
.
2
1
7
3

3
8
S
r
S
t
r
o
n
t
i
u
m
8
7
.
6
2
5
.
6
9
4
9

5
6
B
a
B
a
r
i
u
m
1
3
7
.
3
2
7
5
.
2
1
1
7

7
3
T
a
T
a
n
t
a
l
u
m
1
8
0
.
9
4
7
9
7
.
5
4
9
6

5
4
X
e
X
e
n
o
n
1
3
1
.
2
9
3
1
2
.
1
2
9
8

1
9
K
P
o
t
a
s
s
i
u
m
3
9
.
0
9
8
3
4
.
3
4
0
7

2
0
C
a
C
a
l
c
i
u
m
4
0
.
0
7
8
6
.
1
1
3
2

2
1
S
c
S
c
a
n
d
i
u
m
4
4
.
9
5
5
9
1
0
6
.
5
6
1
5

2
2
T
i
T
i
t
a
n
i
u
m
4
7
.
8
6
7
6
.
8
2
8
1

3
0
Z
n
Z
i
n
c
6
5
.
4
0
9
9
.
3
9
4
2

3
1 G
a
G
a
l
l
i
u
m
6
9
.
7
2
3
5
.
9
9
9
3

3
2 G
e
G
e
r
m
a
n
i
u
m
7
2
.
6
4
7
.
8
9
9
4

3
3
A
s
A
r
s
e
n
i
c
7
4
.
9
2
1
6
0
9
.
7
8
8
6

3
4
S
e
S
e
l
e
n
i
u
m
7
8
.
9
6
9
.
7
5
2
4

3
5
B
r
B
r
o
m
i
n
e
7
9
.
9
0
4
1
1
.
8
1
3
8

3
6
K
r
K
r
y
p
t
o
n
8
3
.
7
9
8
1
3
.
9
9
9
6

2
3
V
V
a
n
a
d
i
u
m
5
0
.
9
4
1
5
6
.
7
4
6
2

2
4
C
r
C
h
r
o
m
i
u
m
5
1
.
9
9
6
1
6
.
7
6
6
5

2
5 M
n
M
a
n
g
a
n
e
s
e
5
4
.
9
3
8
0
4
9
7
.
4
3
4
0

2
6
F
e
I
r
o
n
5
5
.
8
4
5
7
.
9
0
2
4

2
7 C
o
C
o
b
a
l
t
5
8
.
9
3
3
2
0
0
7
.
8
8
1
0

2
8
N
i
N
i
c
k
e
l
5
8
.
6
9
3
4
7
.
6
3
9
8
U
u
b
U
n
u
n
b
i
u
m
(
2
8
5
)
U
u
q
U
n
u
n
q
u
a
d
i
u
m
(
2
8
9
)
U
u
h
U
n
u
n
h
e
x
i
u
m
(
2
9
2
)
N
I
S
T

S
P

9
6
6

(
S
e
p
t
e
m
b
e
r

2
0
0
3
)
P e r i o d
1 6 5 4 3 2 7
F
o
r

a

d
e
s
c
r
i
p
t
i
o
n

o
f

t
h
e

d
a
t
a
,

v
i
s
i
t

p
h
y
s
i
c
s
.
n
i
s
t
.
g
o
v
/
d
a
t
a

2
S
1
/
2
1
s
2
2
s
2
2
S
1
/
2
2
S
1
/
2
[
N
e
]
3
s
2 1
S
0
[
N
e
]
3
s
1
S
0
1
S
0
2
S
1
/
2
1
S
0
2
S
1
/
2
1
S
0
2
S
1
/
2
1
S
0
[
A
r
]
4
s
2
[
A
r
]
4
s
[
K
r
]
5
s
2
[
K
r
]
5
s
[
X
e
]
6
s
2
[
X
e
]
6
s
[
R
n
]
7
s
2
[
R
n
]
7
s
1
G
4
[
X
e
]
4
f
5
d
6
s
2
2
D
3
/
2
3
F
2
2
D
3
/
2
3
F
2
3
F
2
3
F
2
[
A
r
]
3
d
4
s
2
[
A
r
]
3
d
2
4
s
2
[
K
r
]
4
d
5
s
2
[
K
r
]
4
d
2
5
s
2
[
X
e
]
4
f
1
4
5
d
2
6
s
2
[
R
n
]
5
f
1
4
6
d
2
7
s
2
?
4
F
3
/
2
7
S
3
6
D
1
/
2
7
S
3
4
F
3
/
2
5
D
0
[
X
e
]
4
f
1
4
5
d
3
6
s
2
[
X
e
]
4
f
1
4
5
d
4
6
s
2
[
K
r
]
4
d
4
5
s
[
K
r
]
4
d
5
5
s
[
A
r
]
3
d
3
4
s
2
[
A
r
]
3
d
5
4
s
6
S
5
/
2
5
D
4
[
A
r
]
3
d
5
4
s
2
[
A
r
]
3
d
6
4
s
2
6
S
5
/
2
6
S
5
/
2
[
X
e
]
4
f
1
4
5
d
5
6
s
2
[
K
r
]
4
d
5
5
s
2
4
F
9
/
2
[
A
r
]
3
d
7
4
s
2
4
F
9
/
2
[
K
r
]
4
d
8
5
s
3
F
4
2
S
1
/
2
5
F
5
[
K
r
]
4
d
7
5
s
5
D
4
[
X
e
]
4
f
1
4
5
d
6
6
s
2
4
F
9
/
2
[
X
e
]
4
f
1
4
5
d
7
6
s
2
2
S
1
/
2
[
K
r
]
4
d
1
0
5
s
1
S
0
[
K
r
]
4
d
1
0 3
D
3
[
X
e
]
4
f
1
4
5
d
9
6
s
2
S
1
/
2
[
X
e
]
4
f
1
4
5
d
1
0
6
s
1
S
0
2
P
1
/
2
1
S
0
[
K
r
]
4
d
1
0
5
s
2
[
K
r
]
4
d
1
0
5
s
2
5
p
[
X
e
]
4
f
1
4
5
d
1
0
6
s
2
1
S
0
[
H
g
]
6
p
2
P
1
/
2
1
s
2
2
s
2
2
p
1
S
0
1
s
2
3
P
0
1
s
2
2
s
2
2
p
2
4
S
3
/
2
1
s
2
2
s
2
2
p
3
3
P
2
1
s
2
2
s
2
2
p
4
2
P
3
/
2
1
s
2
2
s
2
2
p
5
1
S
0
1
s
2
2
s
2
2
p
6
2
P
1
/
2
3
P
0
4
S
3
/
2
3
P
2
2
P
3
/
2
1
S
0
3
P
0
4
S
3
/
2
3
P
2
2
P
3
/
2
1
S
0
2
P
1
/
2
3
P
0
4
S
3
/
2
3
P
2
2
P
3
/
2
1
S
0
2
P
1
/
2
3
P
0
4
S
3
/
2
3
P
2
2
P
3
/
2
1
S
0
[
A
r
]
3
d
1
0
4
s
2
4
p
[
A
r
]
3
d
1
0
4
s
2
[
A
r
]
3
d
8
4
s
2
[
A
r
]
3
d
1
0
4
s
[
A
r
]
3
d
1
0
4
s
2
4
p
2
[
K
r
]
4
d
1
0
5
s
2
5
p
2
[
A
r
]
3
d
1
0
4
s
2
4
p
3
[
K
r
]
4
d
1
0
5
s
2
5
p
3
[
A
r
]
3
d
1
0
4
s
2
4
p
4
[
K
r
]
4
d
1
0
5
s
2
5
p
4
[
A
r
]
3
d
1
0
4
s
2
4
p
5
[
K
r
]
4
d
1
0
5
s
2
5
p
5
[
A
r
]
3
d
1
0
4
s
2
4
p
6
[
K
r
]
4
d
1
0
5
s
2
5
p
6
[
H
g
]
6
p
2
[
H
g
]
6
p
3
[
H
g
]
6
p
4
[
H
g
]
6
p
5
[
H
g
]
6
p
6
2
D
3
/
2
[
X
e
]
4
f
1
4
5
d
6
s
2
1
S
0
[
X
e
]
4
f
1
4
6
s
2
[
N
e
]
3
s
2
3
p
[
N
e
]
3
s
2
3
p
2
[
N
e
]
3
s
2
3
p
3
[
N
e
]
3
s
2
3
p
4
[
N
e
]
3
s
2
3
p
5
[
N
e
]
3
s
2
3
p
6
[
X
e
]
4
f
1
3
6
s
2
[
X
e
]
4
f
1
2
6
s
2
[
X
e
]
4
f
1
1
6
s
2
[
X
e
]
4
f
1
0
6
s
2
[
X
e
]
4
f
9
6
s
2
[
X
e
]
4
f
7
5
d
6
s
2
[
X
e
]
4
f
7
6
s
2
[
X
e
]
4
f
6
6
s
2
[
X
e
]
4
f
5
6
s
2
[
X
e
]
4
f
4
6
s
2
[
X
e
]
4
f
3
6
s
2
[
X
e
]
4
f
5
d
6
s
2
[
X
e
]
5
d
6
s
2
[
R
n
]
5
f
1
4
7
s
2
7
p
?
[
R
n
]
5
f
1
4
7
s
2
[
R
n
]
5
f
1
3
7
s
2
[
R
n
]
5
f
1
2
7
s
2
[
R
n
]
5
f
1
1
7
s
2
[
R
n
]
5
f
1
0
7
s
2
[
R
n
]
5
f
9
7
s
2
[
R
n
]
5
f
7
6
d
7
s
2
[
R
n
]
5
f
7
7
s
2
[
R
n
]
5
f
6
7
s
2
[
R
n
]
5
f
4
6
d
7
s
2
[
R
n
]
5
f
3
6
d
7
s
2
[
R
n
]
5
f
2
6
d
7
s
2
[
R
n
]
6
d
2
7
s
2
[
R
n
]
6
d
7
s
2
2
D
3
/
2
1
G
4
2
D
3
/
2
3
F
2
4
I
9
/
2
5
I
4
5
I
8
4
I
1
5
/
2
4
I
1
5
/
2
5
I
8
6
H
5
/
2
7
F
0
8
S
7
/
2
9
D
2
6
H
1
5
/
2
2
F
7
/
2
3
H
6
2
P
1
/
2
1
S
0
2
F
7
/
2
3
H
6
9
D
2
6
H
1
5
/
2
7
F
0
8
S
7
/
2
6
L
1
1
/
2
5
L
6
4
K
1
1
/
2
L a n t h a n i d e s A c t i n i d e s
G
r
o
u
p
1
I
A
2
I
I
A
3
I
I
I
B
4
I
V
B
5
V
B
6
V
I
B
7
V
I
I
B
9
V
I
I
I
8
1
0
1
1
I
B
1
2
I
I
B
1
3
I
I
I
A
1
4
I
V
A
S
t
a
n
d
a
r
d

R
e
f
e
r
e
n
c
e
D
a
t
a

G
r
o
u
p
w
w
w
.
n
i
s
t
.
g
o
v
/
s
r
d
P
h
y
s
i
c
s
L
a
b
o
r
a
t
o
r
y
p
h
y
s
i
c
s
.
n
i
s
t
.
g
o
v
1
5
V
A
1
6
V
I
A
1
7
V
I
I
A
1
8
V
I
I
I
A
F
r
e
q
u
e
n
t
l
y

u
s
e
d

f
u
n
d
a
m
e
n
t
a
l

p
h
y
s
i
c
a
l

c
o
n
s
t
a
n
t
s
1

s
e
c
o
n
d

=

9

1
9
2

6
3
1

7
7
0

p
e
r
i
o
d
s

o
f

r
a
d
i
a
t
i
o
n

c
o
r
r
e
s
p
o
n
d
i
n
g

t
o

t
h
e

t
r
a
n
s
i
t
i
o
n
s
p
e
e
d

o
f

l
i
g
h
t

i
n

v
a
c
u
u
m
2
9
9

7
9
2

4
5
8

m

s
1
P
l
a
n
c
k

c
o
n
s
t
a
n
t
6
.
6
2
6
1

1
0
3
4

J

s

e
l
e
m
e
n
t
a
r
y

c
h
a
r
g
e
e
l
e
c
t
r
o
n

m
a
s
s
p
r
o
t
o
n

m
a
s
s
f
i
n
e
-
s
t
r
u
c
t
u
r
e

c
o
n
s
t
a
n
t
1
/
1
3
7
.
0
3
6
R
y
d
b
e
r
g

c
o
n
s
t
a
n
t
1
0

9
7
3

7
3
2

m
1
B
o
l
t
z
m
a
n
n

c
o
n
s
t
a
n
t
1
.
3
8
0
7

1
0
2
3

J

K

1
c h e m
e
k
F
o
r

t
h
e

m
o
s
t

a
c
c
u
r
a
t
e

v
a
l
u
e
s

o
f

t
h
e
s
e

a
n
d

o
t
h
e
r

c
o
n
s
t
a
n
t
s
,

v
i
s
i
t

p
h
y
s
i
c
s
.
n
i
s
t
.
g
o
v
/
c
o
n
s
t
a
n
t
s
b
e
t
w
e
e
n

t
h
e

t
w
o

h
y
p
e
r
f
i
n
e

l
e
v
e
l
s

o
f

t
h
e

g
r
o
u
n
d

s
t
a
t
e

o
f

1
3
3
C
s

(
e
x
a
c
t
)
0
.
5
1
1
0

M
e
V
1
3
.
6
0
5
7

e
V
R R
c
R
h
c
(
/
2
)
m
e
c
2
m
p
1
.
6
0
2
2

1
0
1
9

C

9
.
1
0
9
4

1
0
3
1

k
g
1
.
6
7
2
6

1
0
2
7

k
g

3
.
2
8
9

8
4
2

1
0
1
5

H
z
Bibliography
[1] J. R. Taylor, An Introduction to Error Analysis. University Science Books (Sausalito, Ca),
2nd ed., 1997.
[2] J. I. Pfeer and S. Nir, Modern Physics. Imperial College Press (London), 1st ed., 2000.
[3] P. LeClair, 2009. unpublished work.
[4] R. S. Peterson, Experimental ray spectroscopy and investigations of environmental ra-
dioactivity, 1996. Hard copy available in the laboratory; soft copy available on spectrometer
computer.
[5] M. Planck, ber das Gesetz der Energieverteilung im Normalspektrum, Ann. Phys., vol. 309,
p. 533, 1901. English translation: http://dbhs.wvusd.k12.ca.us/webdocs/Chem-History/
Planck-1901/Planck-1901.html.
[6] A. Einstein, ber einen die Erzeugung and Verwandlung des Lichtes betreenden heuristis-
chen Gesichtspunkt, Ann. d. Physik, vol. 17, p. 132, 1905.
[7] R. Millikan, A direct photoelectric determination of Plancks h, Phys. Rev., vol. 7, p. 355,
1916.
[8] P. Mohr, B. Taylor, and D. Newell, CODATA recommended values of the fundamental phys-
ical constants: 2006, Rev. Mod. Phys., vol. 80, p. 633, 2008. See also the NIST database:
http://physics.nist.gov/cgi-bin/cuu/Value?h.
[9] D. Evans, Measurement of Boltzmanns constant, Phys. Educ., vol. 21, pp. 296299, 1986.
[10] F. Inman and C. Miller, The measurement of e/k in the introductory physics laboratory,
Am. J. Phys., vol. 41, pp. 34951, 1973.
[11] F. Inman, A simple laboratory experiment to measure e/k, The Physics Teacher, vol. 43,
pp. 2728, 2005.
[12] M. E. M. da Piedade and M. N. Berberan-Santos, Atomic emission spectra using a UV-Vis
spectrophotometer and an optical ber guided light source, J. Chem. Ed., vol. 75, pp. 1013
1017, 1998.
229
230 BIBLIOGRAPHY
[13] O. Optics. Operating instructions manual for USB-650 and USB-4000 spectrometers and
http://www.oceanoptics.com.
[14] CRC Handbook of Chemistry and Physics, online edition. http://www.hbcpnetbase.com.
[15] H. G. Kuhn, Atomic Spectra. Longmans (London), 1964.
[16] V. Kondratyev, The Structure of Atoms and Molecules. Dover (New York), 1965.
[17] J. Hubbel and S. Seltzer, Tables of x-ray mass attenuation coecients and mass energy-
absorption coecients. NISTIR 5632. Available online at http://physics.nist.gov/
PhysRefData/XrayMassCoef/cover.html.
[18] G. Audi and A. H. Wapstra, Masses, Q-values and nucleon separation energies: The 1995
update to the atomic mass evaluation, Nuclear Physics, vol. A595, pp. 409480, 1995. http:
//nucleardata.nuclear.lu.se/nucleardata/toi/.
[19] National nuclear data center. http://www.nndc.bnl.gov/index.jsp.
[20] Example gamma ray spectrum and decay schematics for isotopes. http://www.
radiochemistry.org/periodictable/gamma_spectra/.
[21] C. Kittel, Introduction to Solid State Physics. Wiley, 7th ed., 1995.
[22] L. Solymar and D. Walsh, Electrical Properties of Materials. Oxford, 7th ed., 2004.
[23] R. Dragoset, A. Musgrove, C. Clark, and W. Martin. Work of the US government excluded
from copyright, http://www.nist.gov/physlab/data/periodic.cfm.

ph255 Lab Manual PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

ph255 Lab Manual PDF

Hochgeladen von

Copyright:

Verfügbare Formate

PH255: Modern Physics Laboratory

is the energy of the gamma ray in MeV. For gamma rays

Poisson distribution (5.4)

(x) as the probability for measuring x decays per

x. This is a very important result for analyzing experimental data. As a practical

N. Based on your observations, you can

with x. Calculate and comment on the numerical

subsequently (2.55 min half

subsequently (2.55 min half

0 100 200 300 400

with respect to the Helmholtz coils. Thus, you can rotate

green 549.0 546.1

The yellow line is a doublet of 578 and 580 nm.

=2 and n runs from 3 upward. From http://en.wikipedia.org/wiki/Hydrogen_spectrum.

=2 is the same for every line, we

656.285 180 Balmer 3 2

Spin-orbit split lines, 2p

(o) denotes ortho-helium, (p) denotes para-helium

Type Peak (nm) FWHM

Wavelengths should be veried for your particular LEDs.

FWHM = full width at half maximum

oset, essentially no light passes

) component to the desired (0

, they are said to be crossed.

, we see that the

and record the resistance

to ensure that the minima and maxima

and the second to 90

. Place a third polarizer between the two

=2 and n runs from 3 upward. From http://hyperphysics.phy-astr.gsu.edu/.

=2 is the same for every line, we

. If the 0 of the inner

and compare your estimate with the tabulated value.

656.2852 180 Balmer 3 2

Spin-orbit split lines, 2p

with respect to the rows of atoms, we should expect a dierent

j along a dierent direction. The indices of signify which component of

, with the solid sphere of radius R at

. Equation 15.89 is readily parseable: P

results from uid ow

. The electron is ejected with

. This maximum energy transfer is thus:

will result in electrons

> 1.022 MeV

Hz). Integrated circuit devices, such as opera-

R, and a noise current

N, and this is at the heart of signal averaging: repeated measurements

N), the mea-

is small compared to its internal resistance),

, our current sink,"

due to a current through both probes I

Das könnte Ihnen auch gefallen